Self-assembled multi-component catenanes: The effect of multivalency and cooperativity on structure and stability

Article abstract of DOI:10.1021/ja302347q

Using dynamic combinatorial chemistry, mixtures of dipeptide monomers were combined to probe how the structural elements of a family of self-assembled [2]-catenanes affect their equilibrium stability versus competing non-catenated structures. Of particula

Full text of DOI:10.1021/ja302347q

Article
pubs.acs.org/JACS
Self-Assembled Multi-Component Catenanes: The Effect of
Multivalency and Cooperativity on Structure and Stability
Mee-Kyung Chung,† Stephen J. Lee,‡ Marcey L. Waters,*,† and Michel R. Gagne*,†
́
†Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290, United States
‡U.S. Army Research Office, P.O. Box 12211, Research Triangle Park, North Carolina 27709, United States
S
* Supporting Information
ABSTRACT: Using dynamic combinatorial chemistry, mix-
tures of dipeptide monomers were combined to probe how the
structural elements of a family of self-assembled [2]-catenanes
affect their equilibrium stability versus competing non-
catenated structures. Of particular interest were experiments
to target the effects of CH−π interactions, inter-ring hydrogen
bonds, and β-turn types on [2]-catenane energetics. The non-
variant core of the [2]-catenane was shown only to adopt type
II′ and type VIII turns at the β-2 and β-4 positions,
respectively. Monomers were designed to delineate how
these factors contribute to [2]-catenane equilibrium speciation/stability. Dipeptide turn adaptation studies, including three-
component dynamic self-assembly experiments, suggested that stability losses are localized to the mutated sites, and that the turn
types for the core β-2 and β-4 positions, type II′ and type VIII, respectively, cannot be modified. Mutagenesis studies on the core
Aib residue involved in a seemingly key CH−π−CH sandwich reported on how CH−π interactions and inter-ring hydrogen
bonds affect stability. The interacting methyl group of Aib could be replaced with a range of alkyl and aryl substituents with
monotonic affects on stability, though polar heteroatoms were disproportionately destabilizing. The importance of a key cross-
ring H-bond was also probed by examining an Aib for L-Pro variant. Inductive affects and the effect of CH donor multiplicity on
the core proline−π interaction also demonstrated that electronegative substituents and the number of CH donors can enhance
the effectiveness of a CH−π interaction. These data were interpreted using a cooperative binding model wherein multiple non-
covalent interactions create a web of interdependent interactions. In some cases, changes to a component of the web lead to
compensating effects in the linked interactions, while in others, the perturbations create a cascade of destabilizing interactions
that lead to disproportionate losses in stability.
easily varied interactions creates a new tool for the study of
multivalency and cooperative binding in complex systems.
In our previous paper,3 we described a phenomenon wherein
mixtures of dipeptides self-assemble under dynamic conditions
into stereochemically and constitutionally precise [2]-cate-
nanes, wherein one 56-membered macrocyclic ring interlocks
with a second ring to give highly ordered structures that are
intermediate in size and complexity between typical molecular
recognition and biological systems. Despite their large size (>3
kDa), these compounds could be precisely characterized by
high-resolution techniques that include X-ray crystallography,
NMR, and mass spectrometry. These tools revealed that the
[2]-catenanes are assembled from a complex web of molecular
interactions that include both intra- and inter-macrocycle
hydrogen bonds, a broad array of edge-to-face and slipped face-
to-face π−π interactions, and seemingly critical CH−π
interactions. Other defining structural features include the
utilization of β-turns, which, at least in the core, are highly
conserved and adopt biomimetic turn types (β-2 is type II′ and
INTRODUCTION
■
Multivalency and cooperative binding are phenomena that
occur throughout the genre of molecular recognition, though
the latter is principally observed in the biological context.1
While many molecular systems participate in multivalency, true
cooperativity, i.e., binding strengths that are greater than the
sum of the parts, is rare for synthetic systems.2 One contributor
to the chasm between synthetic and biological systems is the
inherently small size and the lack of complexity of synthetic
systems that tend to be amenable to study by traditional
spectroscopic and/or thermochemical means. The process of
“designing” a molecular recognition system additionally creates
a bias toward the molecular interactions that were used to
create the structure’s desired recognition properties. This heavy
hand overwhelms the subtle factors that could conspire to
create cooperativity and additionally distances the results from
the biological context where they are most often observed. The
study of multivalency and cooperative binding is therefore
challenged by the constraints of being “designed” and the
technical difficulties of studying/manipulating complex (large)
systems known to display cooperativity. The generation of
spontaneously assembled systems held together with multiple,
Received: March 9, 2012
Published: June 11, 2012
© 2012 American Chemical Society
11430
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
β-4 is type VIII). Each of these non-covalent interactions and
the β-turn secondary structures are common elements in
protein structures.
examine multivalency and perhaps even cooperativity in a
regime of significant structural complexity. As such, this system
enables one to begin revealing how the nature of molecular
interactions can be modulated by other structural elements.
In the [2]-catenanes characterized in our previous paper,3
one invariant region of the structure is the “core”, where the
two conserved aryl glycine units seed a network of CH−π
interactions and inter-macrocycle hydrogen bonds. The
absolute requirement for an aryl glycine-containing monomer
in the dynamic self-assembly reaction (DSA), and the
observation that each arylGly unit engages in CH−π
interactions on both π−faces, led to the hypothesis that these
interactions, expressed through the CH−π−CH sandwich, were
acting to template the synthesis and stabilize the catenane.
Relatively weak CH−π interactions are not well studied,
despite their ubiquity and influence over important chemical
phenomena, which include small-molecule conformation,4
chiral discrimination,5 assembly of host−guest complexes,4a,6
molecular recognition,7 selectivity of organic reactions,6a,8
crystal packing,6a,9 bioconjugates,4 and protein−protein inter-
actions involving polyproline helices.10 The well-characterized
nature of the catenane and the modular, thermodynamically
controlled self-assembly/synthesis protocol suggested an
experimental strategy for probing the importance and the
properties of the CH−π interactions present in the self-
assembled [2]-catenanes. The desire to examine the CH−π
interaction in a complex setting was stimulated by the
constraints of small-molecule studies, which can measure the
strength of the CH−π interaction in isolation, and computa-
tional studies, which can probe the CH−π interaction in simple
systems. Studies capable of manipulating this interaction in a
complex, multivalent scheme that bridge the gap to the
biological context are scant.
EXPERIMENTAL SECTION
■
General Methods. Chemicals were purchased from Aldrich, Fisher
Scientific, and Chem-Impex International, Inc. and used as received.
1H and 13C NMR spectra of monomers were recorded on a Bruker
DRX 500 spectrometer and processed using TOPSPIN Bruker NMR
software (V. 3.0). High-resolution mass spectra were obtained on an
Agilent Accurate LC-TOF mass spectrometer (Agilent Series 6220)
operating in positive-ion mode with an electrospray ionization source
(fragmentor = 175 V). The data were analyzed using Agilent
MassHunter Workstation Software, Qualitative Analysis (V.
B.02.00). HPLC analysis was performed on a Hewlett-Packard Series
1100 instrument, using a Halo-C18 column (4.6 × 150 mm, 2.7 μm)
with gradient elution (methanol/water) at a flow rate of 0.45 mL/min
and at 55 °C. The injection volume for a 5 mM dynamic self-assembly
(DSA) reaction was typically 3.5 μL. UV absorbance chromatograms
were recorded at wavelengths of 220 and 289 nm. The data were
analyzed using the Agilent Chemstation software. Accurate Mass LC-
TOF analyses of DSA reactions were performed on an Agilent Series
1200 LC instrument, using a Halo-C18 column (2.1 × 50 mm, 2.7
μm) with gradient elution (methanol/water containing 0.1% formic
acid) at a flow rate of 0.5 mL/min and at 50 °C. The eluent was
analyzed by an Accurate Mass LC-TOF mass spectrometer (Agilent
Series 6220) in positive-ion mode with an electrospray ionization
source (fragmentor = 375 V). The data were analyzed using Agilent
MassHunter Workstation Software, Qualitative Analysis (V. B.02.00).
Self-Assembly of [2]-Catenanes for HPLC and LC-TOF
Analyses. Five millimolar DSA solutions were prepared on a 1 mL
scale. D- or D,L-1x (5 μmol) and L,L-y2 or L,L-y2Ph (5 μmol) were
separately dissolved in 1 mL of a mixture of MeCN and CHCl3 (1:3),
and 0.5 mL of the D- or D,L-1x solution was mixed with 0.5 mL of the
L,L-y2 or L,L-y2Ph solution and trifluoroacetic acid (50 equiv). The
solutions were allowed to sit for several days until a steady state was
reached (typically 7−10 days) prior to LC analyses.
Speciation and Quantification of [2]-Catenanes by HPLC
and LC-TOF. The speciation and characterization of [2]-catenanes in
the DSA solution were performed by HPLC and LC-TOF analyses.
For HPLC analyses, 3.5 μL of DSA solution was injected onto a Halo-
C18 column (4.6 × 150 mm, 2.7 μm) at a flow rate of 0.45 mL/min at
55 °C. To remove solvent and TFA, a short linear gradient (35−46%
MeOH/H2O over 4 min) was used. A second gradient composed of
two linear gradients (46−66% MeOH/H2O over 26 min and
consecutive 66% MeOH/H2O to 100% MeOH over 12 min) was
used to separate non-catenane species, and then 100% MeOH was
subsequently used for an additional 12 min for [2]-catenane
speciation. The [2]-catenanes in the DSA solution were quantified
from their UV trace areas (289 nm) and compared to a calibration
curve generated from known concentrations of several isolated
catenanes and their corresponding UV trace areas (289 nm). The
[2]-catenanes in the DSA solution were identified by LC-TOF
analyses in positive-ion mode with an electrospray ionization source
(fragmentor = 375 V, Vcap = 3500 V, mass range = 100−3200 m/z).
A 1.0 μL DSA solution was injected onto a Halo-C18 column (2.1 ×
50 mm, 2.7 μm) at a flow rate of 0.50 mL/min at 50 °C. To remove
non-catenane species, solvent, and TFA, a short gradient (60−70%
MeOH/H2O containing 0.1% formic acid) was used. A second
gradient (70−90% MeOH/H2O containing 0.1% formic acid over 7
min) and a subsequent 90% MeOH/H2O (0.1% formic acid) mobile
phase were utilized for an additional 4 min for catenane speciation. See
Supporting Information-2 for HPLC traces of DSA solutions and LC-
TOF analyses of the generated catenanes.
With the goal of delineating a structure/stability relationship,
we report the outcome of our investigations to manipulate the
[2]-catenane’s defining CH−π interactions and subsequently its
core β-turns. The previous paper3 provided a structural
blueprint for designing modified monomers to perturb the
strength of the CH−π interaction, as determined from the
relative stability of the folded [2]-catenane versus competing
non-catenated species. The nature of these experiments
required careful controls to support the notion that the
examined subtle perturbations were principally affecting the
highly structured and folded catenanes rather than the non-
catenated cyclic oligomers.
These mutation studies indicate that a minimum number of
CH−π interactions are necessary, but not sufficient for the self-
assembly of these catenanes. Mutations in the CH−π
participating core β-turns (β-2 and β-4) revealed that the
conserved turn structures were critical to the stability of the
[2]-catenanes. Structure−stability correlations showed the
CH−π interactions to be part of an interconnected network
of non-covalent interactions that encompass the entire
structure. Semi-quantitative measurements of stability indicate
that even small structural perturbations can become magnified
or attenuated in this network. For example, the excision of two
CH3 groups involved in CH−π interactions causes a >1000-fold
drop in catenane stability. As will be discussed, the source of
this loss is a combination of CH−π loss, a change in β-turn
conformational preference, and a concomitant loss of highly
conserved inter-ring hydrogen bonds. In other mutations, the
expected drop in stability was attenuated, and metrical evidence
points to compensating structural reorganizations which restore
some of the lost stability. These experiments thus enable one to
11431
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Among these non-covalent interactions, the most intriguing
were the CH−π interactions which grip the naphthyl ring of
naphthylglycine between a proline and an Aib residue (Figure
1b) (Aib = 1-aminoisobutyric acid, or α-methyl-alanine). The
rigidity of the peptide loops and the core structure were probed
through H/D exchange experiments in protic D-solvents
(CD3OD, D2O/pyridine). The exchangeable amide and
hydrazone NH groups revealed a variety of chemical environ-
ments, including those in the outer loops which exchange
readily, and those flanking the core naphthyl, which were much
slower. Two key NH’s forming inter-ring hydrogen bonds in
the core (β-2 and β-4 turns of the catenane H3a; see turn
designations in Figure 1b) resisted exchange even after multiple
weeks in an aggressive H/D exchange solvent (D2O/pyridine).
In other examined catenanes, this H/D exchange eventually
occurred but much more slowly than in the outer loops. As fast
exchange was taken to reflect structural flexibility and
breathability, these observations suggest that the core is
rigidified/stabilized by the CH−π interactions, which sandwich
the naphthyl rings and pairs of inter-ring hydrogen bonds.
X-ray structure analysis3 additionally revealed that the core of
each catenane is composed of a type II′ β-2 turn and a type VIII
β-4 turn. In contrast, the peripheral turns (β-1 and β-3) can
adopt either type I′, type I, or type II′ β-turns. The invariability
of the core suggests that these β-turn types may also play a role
in facilitating the above core interactions.
RESULTS
■
1. [2]-Catenane Generation. Under dynamic self-
assembly (DSA) conditions, mixtures of two dipeptide
monomers that are epimeric at proline (D-1a and L,L-H2 or
H2Ph) self-assemble into octameric [2]-catenanes (Scheme 1)11
Scheme 1. [2]-Catenane Self-Assembly from Two Dipeptide
Monomers (D-1a and L,L-H2 or 2Ph)
H
Given the notion that the CH−π interaction was the key
differentiating recognition element, we first sought to gain
insight by perturbing the β-2 CH donor (Scheme 2b) and
H
in a diastereoselective and constitutionally selective fashion. X-
ray structure analysis3 showed that the structure contains a
dense core from which extend loops terminating in β-turns.
As shown in Figure 1a, the structure is flower-like and
assembled from two interlocked saddle-shaped loops where
Scheme 2. Illustration of the CH−π−CH Sandwich in 3a
Showing CH−π Interactions (a) between the Naphthyl and
L-Pro Moiety (β-4 CH−π) and (b) between the Naphthyl
and a Methyl of the Aib Residue (β-2 CH−π)
Figure 1. (a) X-ray structure of H3a displaying two identical
interlocked tetramers. (b) View from the bottom-side of (a) showing
CH−π interactions between the naphthyl (space-filled shape) and L-
Pro moiety and between the naphthyl and a methyl group of the Aib
residue. All hydrogen atoms except those participating in the CH−π−
CH sandwiches are omitted for clarity.
determining how the [2]-catenane responded. To this end, a
series of methyl deletion experiments were designed to
manipulate the Aib residue of β-2. Since this group originates
from the D-Pro-Aib unit, this experiment was accomplished by
employing the D-Pro-L-Ala, D-Pro-D-Ala, or D-Pro-Gly mono-
mers for the DSA reaction with monomer L,L-H2 (Table 1). As
described in Scheme 2b, this series of monomers remove the
non-interacting methyl group (Me2) in the D-Pro-L-Ala case,
the interacting methyl group (Me1) in the D-Pro-D-Ala case,
and both groups in D-Pro-Gly scenario. Not surprisingly, the D-
Pro-L-Ala situation (entry 2) gave only a slightly lowered
concentration of [2]-catenanes relative to Aib (50 vs 70%), as
this methyl deletion combination should leave the CH−π
interaction largely untouched. The D-Pro-D-Ala scenario (entry
each macrocycle is formed from three units of 1 and one unit of
2, i.e., [1321]2. X-ray and NMR structural investigation of
catenane H3a revealed that its dense self-complementary
structure is stabilized by a combination of six intra- and six
inter-macrocyclic hydrogen bonds, numerous face-to-face and
edge-to-face aromatic interactions, and the previously men-
tioned CH−π interactions. These interactions were commonly
found in all examined catenanes.3
11432
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
only accommodates a type II′ β-2 and a type VIII β-4 turn, it
become necessary to determine whether the adaptability of the
dipeptides to these turn types was a prerequisite for catenane
formation. We first address this issue for the β-2 turn in detail
using the D-Pro-D-Ala monomer, and then we present the result
of additional perturbations to the β-2 face of the aryl ring,
simultaneously considering the natural β-turn preferences of
the monomer variations (Scheme 3). This modulation sought
Table 1. [2]-Catenanes Generation from D-, D,L-, or D,D-1x (x
= a, b, v) and L,L-H2
Scheme 3. [2]-Catenane Self-Assembly from Pairs of
Hydrazine/Acetal Monomers
a
Yields obtained from the UV trace of the corresponding catenane in
each DSA solution at day 7−9 (steady state) using a calibration curve.
The calibration curve was obtained from the HPLC-UV area (289 nm)
of several known concentrations of pure catenanes; details in SI1
Figure 1. The values in parentheses were obtained from 3:1 (D:L)
b
biased DSA solutions. Other constitutional isomers of 3x and 4x.
c
Only trace amounts of [2]-catenanes were generated (≪ 1%).
3), however, was unexpected, as it led to a large decrease
(>1000-fold) in catenanes. While on the surface this suggested
that the CH3−π interactions (there are two C2-related
positions) were stronger than expected from model studies
and calculations,4a,12 this interpretation was immediately
questionable, as the D-Pro-Gly case (entry 4), which also
lacks the interacting Me1 group, forms a significant quantity of
[2]-catenanes (33%). These data require an explanation that
extends beyond the simple CH−π interaction mode.
A likely source for this trend became apparent when the
inherent β-turn preference of the core units was considered, as
the D-Pro-Aib, D-Pro-D-Ala, D-Pro-L-Ala, and D-Pro-Gly units
occupy the invariant β-2 location in the core. Revealing was the
fact that D-Pro-L-Ala, D-Pro-Gly, and D-Pro-Aib each prefer type
II′ β-turns, while D-Pro-D-Ala prefers a type I′ β-turn.13 That is,
the three monomers preferring a type II′ turn form significant
[2]-catenane, while the single case which does not is highly
destabilized. The lowered quantity of [2]-catenanes for D-Pro-
Gly relative to entries 1 and 2 indicates that, while the CH−π
interaction is apparently stabilizing, this effect is subordinate to
the ability to adopt a β-2 type II′ turn.
While the stability trends for entries 1, 2, and 4 suggest
stability perturbations through CH−π effects, these data
stimulated a broader investigation examining β-turn prefer-
ences, predisposition to key hydrogen bonds, and CH−π
interactions. This study was broken down into two main
components, each focusing on a different face of the aryl ring of
the CH−π−CH sandwich (β-2 and β-4). Since the five X-ray
structures and additional NMR spectra suggest that the core
to vary the structure of the β-2 CH donor to determine how
catenane stability was affected, vis-a-vis its equilibrium
̀
concentration. This was achieved by substituting the D-Pro-
Aib unit for a variety of D-Pro-L and D-Pro-cyclic dipeptide
structures. Since D-Pro-L-X and D-Pro-cyclo-X dipeptides can
adopt type II′ turns,14 these variants were thought to principally
vary the β-2 CH donor of the CH−π−CH sandwich. The β-4
CH donor component of the CH−π−CH sandwich and the
turn propensities were also examined through proline variants
and the stereoisomeric dipeptides of L-Pro-L-arylGly (H2 and
H2Ph), respectively.
11433
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
The effect of the above-described mutations on catenane
formation/stability was assessed by determining the equilibrium
speciation of the DSA solutions. Typical compositions included
cyclic dimers through hexamers (pentamers and hexamers were
present in low abundance), with each n-mer being composed of
multiple constitutional isomers of the two monomers. HPLC
was used to monitor the approach to the steady state, which
varied for different L-monomers.15 In most cases octameric [2]-
catenanes are visible within 5 min and a significant amount of
catenane accumulates within a day (SI1 Figure 2). Notably, the
tetrameric macrocycle corresponding to one half of 3x (i.e.,
[1321]) is barely detectable, and one of the most abundant
cyclic species at early time, a trimer subunit of 3x composed of
two units of 1 and one unit of 2 ([1221]), substantially
decreases as the cantenane concentrations increase.
To compare [2]-catenane yields across the series, [2]-
catenane concentrations were determined from the HPLC−UV
trace (289 nm) and calibration curves established for several
purified [2]-catenanes; compound-to-compound differences in
extinction coefficient were negligible (SI1 Figure 1). In several
instances a precipitate was noted at the 3−5 day time point of
the DSA reaction. Even when this occurred, the catenane
concentration in the supernatant either continuously increased
to a steady state, which then began to slowly drop after 15 days,
or reached a steady state before the onset of the precipitation.
These non-equilibrium situations are noted.
By quantifying the amount of catenane in a DSA reaction,
one measures how its free energy compares to that of the other
components of the mixture. In addition to affecting the
catenane’s energy, varying the DSA components could also
perturb the relative energy of the other components of the
equilibrium mixture (dimers, trimers, etc.), and thus hinder the
comparability of one system to another. To assess the viability
of cross-system comparisons, we carried out full speciation
analyses on several DSA reactions (D-1a/L,L-H2, D,L-1b/L,L-H2,
D,D-1b/L,L-H2, D-1v/L,L-H2, D-1a/L,L-H2Ph, D,D-1b/L,L-H2Ph, and
D-1v/L,L-H2Ph in Scheme 3) and looked for any unusually stable
(or unstable) cyclic oligomers that could artificially lower (or
raise) the amount of catenanes (see representative cases in SI1
Figure 3−5). These experiments demonstrate that the
dimers:trimers:tetramers ratio does not change significantly
from system to system, and suggest that the structural changes
being explored herein do not significantly affect the relative
energy of the non-catenated, less structured, cyclic oligomers
present in the DSA solution. With the caveat that this analysis
does not rigorously rule out subtle effects, we report both
absolute and relative (to the other DSA components)
concentrations of the [2]-catenanes and interpret these data
to reflect [2]-catenane stability.
2. β-2 Turn and [2]-Catenane Generation. 2.1. Turn
Type. X-ray structure analysis revealed that all examined
catenanes adopt a type II′ β-2 turn in the core, obviously
suggesting that this turn is most suited to a stable [2]-catenane.
The combination of a type II′ β-2 and a type VIII β-4 creates a
core structure wherein two key interactions can take place: (1)
CH−π interactions between the X residue of the D-Pro-X unit
and the aryl residue of the L-Pro-L-arylGly unit and (2) two
inter-ring H-bonds between two β-turn backbones (see red-
boxed interactions in Figure 2). As mentioned previously, the
preferred β-turn of the D-Pro-D-Ala dipeptide is type I′.
Superimposing this mismatched turn geometry onto the
structure of H3a reveals several destabilizing interactions. In
the type I′ turn the amide is rotated nearly 180°,16 which
Figure 2. X-ray structure of H3a displaying the inter-ring hydrogen
bonds (black dot-lines). The red box indicates a CH−π interaction
(black up-down arrow) and inter-ring hydrogen bonds between the β-
2 and β-4 backbones. The blue circle indicates the loss of the inter-ring
hydrogen bond and gain of the repulsive interaction caused by a
proximate carbonyl group in the β-2 turn when a type I′ turn instead of
a type II′ turn is accommodated in the β-2 turn. All hydrogen atoms
except those forming inter-ring hydrogen bonds and the CH−π
interaction (red box) are omitted for clarity.
removes the inter-ring hydrogen bond while simultaneously
creating a repulsive carbonyl−carbonyl interaction (see the
shadow interactions in Figure 2). This repulsive term,
combined with the loss of two CH3−π interactions (doubled
due to C2 symmetry), is the likely cause for the dramatic loss of
[2]-catenane (>1000-fold drop).17
Interestingly, a similar degree of destabilization can be found
in the difference in binding affinity of vancomycin to the D-Ala-
D-Lac mutant relative to D-Ala-D-Ala (1000-fold drop).18 The
introduction of an ester oxygen in D-Ala-D-Lac causes both the
loss of a hydrogen bond and the introduction of a destabilizing
electrostatic repulsion between a carbonyl and the ester oxygen
in D-Ala-D-Lac. This parallel suggests that the loss of the
hydrogen bond and the additional repulsive interaction
expected for D-Pro-D-Ala may be the primary reasons for
destabilization, with loss of the CH3−π interactions playing a
lesser role. This is supported by the ability of the catenane to
accommodate D-Pro-Gly in the core.
Although the above structural rationale for loss of catenane
using D-Pro-D-Ala is strong, two additional explanations were
considered: (1) an exceptionally stable cyclic oligomer was
being formed, with the concomitant effect of draining the
concentration of [2]-catenane, and (2) the loss of stability was
not due to a specific set of interactions at the β-2 position, but
the monomer simply could not occupy one of the other two
positions (β-1 and β-3) in the structure (despite their
heightened structurally flexibility).
To address scenario 1, a full speciation analysis was carried
out on the non-catenated products for the two experimental
permutations: D-Pro-Aib/L-Pro-L-NaphGly (D-1a/L,L-H2),
which gives [2]-catenane, and D-Pro-D-Ala/L-Pro-L-NaphGly
(D,D-1b/L,L-H2), which does not. The results clearly demon-
strate that no unusually stable or unstable n-mers exist in either
11434
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
experiment (see SI1 Figures 3 and 4). When catenane is
formed, this naturally lessens the total amount of n-mers
remaining at equilibrium, but within this group of cyclics, the
distribution of 2-mers:3-mers:4-mers is unchanged.19 Since no
unusual stabilities were evident in the macrocycles, these data
suggest that the lack of [2]-catenane formation with D-Ala lies
in the [2]-catenane itself.
Scheme 4. Logical [2]-Catenane Combinations from Mixing
Two 1 Monomers (D-1a, D,D-1b) with the L,L-H2 Monomer
Given this evidence, the problem logically reduces to the D-
Ala monomer not being accommodated into the β-1, β-2, and
β-3 positions or combinations thereof. The turn type
destabilization rationale presented above would suggest that
the problem is localized at the β-2 site, but since the data do
not rule out the possibility that the problem lies with its
positioning at β-1, β-3, or both, three monomer experiments
were designed. Outlined in Scheme 4 is the logical consequence
of mixing monomer 2 with two different 1 monomers on the
production of 3,20 one of which is the previously unincorpo-
rated D-Pro-D-Ala (D,D-1b), while the second is the successfully
incorporated D-Pro-Aib (D-1a).
In the event that D-Pro-D-Ala is not incorporated in any of
the three positions (i.e., NO, NO; case 1 in Scheme 4), then
the three monomer experiment would form [2]-catenane
incorporating the D-Pro-Aib monomer exclusively, i.e., [(D-
1a)6(L,L-H2)2], with no catenanes containing D,D-1b being
detected. If, on the other hand, the D-Pro-D-Ala monomer is
incorporated into one or a combination of the β-sites, then
different constitutional isomers are possible, each of which can
be discriminated in an LC-TOF analysis. While the β-1 and β-3
sites are chemically inequivalent, the analysis in Scheme 4 treats
them as either equally probable or improbable, though their
contributions to the constitutional isomerism of mixed [2]-
catenanes is implied. As will be seen below, the logical
differentiation between the β-1 and β-3 sites was not necessary
to explain the data.
If the D-Pro-D-Ala monomer could only occupy the β-2 sites
(i.e., YES, NO; case 3 in Scheme 4), only two mixed catenanes
of 3 would be formed. If the opposite were true and D-Pro-D-
Ala was only excluded from the β-2 turn (i.e., NO, YES; case 2
in Scheme 4), then a significant number of mixed [2]-catenanes
would result. For [2]-catenane 3, this latter scenario would
require that each isomer contain a minimum of two D-Pro-Aib
monomers to occupy the β-2/2′ positions, with the balance of
positions being occupied by logical combinations of D-Pro-Aib
and D-Pro-D-Ala. Conversely, a maximum of four D-Pro-D-Ala
units can be accommodated (case 2 in Scheme 4).21
When D-Pro-D-Ala, D-Pro-Aib, and L,L-H2 were combined in a
1:1:2 ratio, a complex mixture of catenanes was obtained,
though the identity of each compound could be deconvoluted
by LC-TOF (Figure 3). Clear from this analysis was the fact
that each catenane contained at least 2 equiv of L,L-H2
(expected) and no less than 2 equiv of D-Pro-Aib. The diversity
in composition in Figure 3a, therefore, results from the D-Ala
and Aib monomers being statistically distributed into the β-1,
β-1′, β-3, and β-3′ sites, with the β-2 and β-2′ sites being solely
occupied by the Aib monomers.
When the ratio of D-Pro and the L-Pro monomers is changed
to 3:1, that is, D-Pro-D-Ala:D-Pro-Aib:L,L-H2 = 3:3:2, the
catenane compositional diversity is unchanged from Figure
3a, though the absolute intensity of each species increases due
to the D-Pro-bias (SI1 Figure 6a). On the other hand, varying
the ratio of the two D-Pro monomers (keeping the D-Pro:L-Pro
ratio at 1:1) shows that D-Pro-D-Ala-rich cases generate less
catenanes and fewer isomers, while D-Pro-Aib-rich cases
generate comparatively more catenane but fewer isomers (SI1
Figure 6b).
We interpret these data to reflect an absolute requirement for
two L,L-H2 monomers in the β-4/4′ positions and two D-Pro-
Aib monomers in the β-2/2′ positions.22 The D-Pro-D-Ala
monomer can therefore be easily accommodated into the β-1
and β-3 sites, but not the β-2 site. We conclude that the loss of
catenane stability is localized to the β-2/2′ sites, and that the
principal requirement for catenane stability is adopting the
necessary type II′ β-turn. Once the turn is in place, the non-
covalent interactions in Figure 2 function to perturb the
comparative stabilities.
11435
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Table 2. Yields of [2]-Catenanes in the DSA Solutions
Formed from Various D-Monomers and L,L-H2
Figure 3. LC-TOF UV trace (289 nm) of [2]-catenanes at day 7
generated from (a) D-1a:D,D-1b:L,L-H2 = 1:1:2 mixture, (b) D-1a:L,L-H2
= 1:1 mixture, and (c) D,D-1b:L,L-H2 = 1:1 mixture (5 mM in 25%
MeCN/CHCl3, 50 equiv of TFA). Each peak was identified by LC-
TOF analysis.
2.2. Aib Modifications in D-Pro-Aib. The X-ray structure
analysis and ROESY NMR spectrum of H3a3,11 established a
CH−π interaction between the nucleated naphthyl group and
one of the Aib methyl groups (β-2). Since only one CH3 group
was upfield shifted, it implied that [2]-catenane stability might
not be restricted to Aib, as other arrangements should be
capable of creating a productive CH−π interaction, provided
that the preferred β-turn is type II′ (vide supra). The catenanes
generated by the D-Pro-L-Ala monomer (entry 2 in Table 1)
support this hypothesis. Candidate structures for varying the β-
2 CH donor include D-Pro-X and D-Pro-L-X dipeptides, as these
predominantly form type II′ β-turns.13c,23
As shown in Table 2 and SI1 Table 1, all tested D- and D,L-
monomers generated the corresponding [2]-catenanes, with
both aliphatic and aromatic groups being well behaved. In the
D,L-monomers having an aliphatic alanine substituent, a clear
steric trend was noted, with larger groups destabilizing the
structure. Considering the dense nature of this region (Figure
4), the stability response to steric changes (n-Bu vs i-Pr vs tert-
Bu) was surprisingly muted, as even a tert-butyl for methyl
substitution provided a measurable quantity of [2]-catenane
(Table 2, entry 7). Polar substituents (OH or F), however,
were disproportionally detrimental to stability.24 This observa-
tion is readily accommodated by a model which invokes a
repulsive polar−π interaction in the constrained pocket.
D,L-Monomers having an arylGly residue also generated the
corresponding [2]-catenanes (Table 2, entries 8−10; SI1 Table
1, entries 4−6), implying that a stabilizing π−π interaction
between the naphthyl group and the arylGly residue can be
accommodated. One example of this family of [2]-catenanes
was crystallographically characterized3 and shown to adopt a
nonsymmetric solid-state structure featuring a CH−π−π
sandwich in the core, wherein the β-2 face creates π−π
interactions that are either of the slipped π−π or edge-to-face
a
Yields obtained from the UV trace (289 nm) of the corresponding
catenane in each DSA solution at day 7−9 using a calibration curve.
The calibration curve was obtained from the HPLC-UV area (289 nm)
of several known concentrations of pure catenanes; Details in SI1
Figure 1. The values in parentheses were obtained from 3:1 (D:L)
b
biased DSA solutions. Other constitutional isomers of 3x and 4x.
c
Biased DSAs (3:1) led to precipitates and yields are not reported.
d
Precipitation occurred within 3−5 days. The yields of the generated
catenanes were obtained at day 7−8 after filtering the precipitates.
variety. In cases where electronically perturbed phenyl glycines
were utilized, the electron-releasing para-methoxy substituent
gave more catenane than did the p-CF3 substituent (Table 2,
entries 9 vs 10), in apparent contradiction to the Hunter−
Sanders model.25
When cyclic α,α-dialkylated analogues of Aib were used,
outcomes that depended on ring size were obtained, with the
optimal size being a C5 or C6 ring (Table 2, entries 11−14). As
predicted by the respective catenane stabilities, the β-turns of D-
11436
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Figure 4. X-ray structure of H3a displaying the compact structure
between the naphthyl and the Aib. Space-filled shapes are used for
these groups.
Figure 5. Picture superimposing an imaginary L-Pro group instead of
the Aib (β-2′) in the X-ray structure of H3a (the black outlined shape
in the blue box). Since L-Pro group has the N (β-2′) instead of the
amide NH, it induces a structure lacking an inter-ring hydrogen bond
(β-2′(NH)→β-4(CO)) between the β-2 and β-4 backbone. All
hydrogen atoms except those forming inter-ring hydrogen bonds are
omitted for clarity.
Pro-1-amino-1-cycloalkanecarboxylic acids (Ac5c and Ac6c) are
similar to those of D-Pro-Aib.14d,26 Although the reasons are not
obvious, these cyclic monomers skew the catenane composi-
tions toward H4x and the other minor catenanes. Since the
cyclic α,α-dialkylated analogues of Aib populate a more
constrained conformational space at both the mutated and
adjacent positions,14e,27 these modifications apparently com-
pensate for the structural changes that result from replacing a
type II′ for a type I turn at the β-3 or β-1 positions.
As mentioned in section 1, the D-Pro-Gly monomer, which
lacks the methyl group that participates in the CH−π
interaction but favors a type II′ β-turn, creates a significant
quantity of [2]-catenane under the DSA conditions.28 The
structural flexibility noted by the steric trend in this position
suggests that the structure could adjust to fill the void and gain
stabilization through a backbone CH−π interaction. Alter-
natively, the competing cyclic n-mers would somehow have
been disadvantaged, though a speciation analysis suggests that
this is not the case (SI1 Figure 5). A three-component DSA
reaction using a 1:1 mixture of D-Pro-Gly/D-Pro-Aib as the 1
components with L,L-H2 returned results indicating that, like D-
Pro-Aib, D-Pro-Gly has an equal propensity to occupy the β-1,
β-2, and β-3 positions (in contrast to D-Pro-D-Ala; see SI1
Figures 8−10).29 In all respects the D-Pro-Gly monomer is
statistically equivalent to the parent monomer (D-Pro-Aib) and
is thermodynamically viable in all positions. These data
therefore imply that a CH−π interaction on the β-2 face of
the arene is not an absolute requirement for competitive
stability or the structure can significantly reorganize to position
the glycine CH2 group into the CH3−π position.
In an attempt to separate the type II′ β-2 turn requirement
from the subsequent inter-ring hydrogen bond, the D-Pro-L-Pro
monomer was tested. This dipeptide favors a type II′ β-turn30
but lacks the amide NH to complete the inter-ring hydrogen
bond (Figure 5). The ∼10% yield of catenanes (see SI1 Table
1) indicates that a structure lacking two (β-2(NH)→β-4(C
O)) of six inter-ring hydrogen bonds can at least partially
compete with non-catenanted macrocycles, though at a price.
The generation of catenanes with this disadvantage implies that
the otherwise conserved inter-ring hydrogen bond is not an
absolute requirement for catenane stability, as its loss can be
partially offset by the network of CH−π, π−π, and inter-ring
hydrogen-bonding interactions.
3. β-4 Turn and [2]-Catenane Generation. 3.1. Turn
Type. X-ray structure analysis and H/D exchange experiments3
indicated that every β-4 turn in the core is of the type VIII form
and this conformation is maintained in solution. Previous
studies have shown that L-Pro-L-X dipeptide units favor type I
turns,31 which is the form adopted by L-Pro-L-NaphGly when it
is located in the flexible β-3 loop of catenane F4u (F4b in ref 3).
In this position, the naphthyl group is not involved in any
CH−π interactions.
Adopting a type VIII turn in β-4 thus requires the L-Pro-L-
arylGly units to undergo a turn change. A comparative analysis
of the standard turn dihedral angles (type I, Φi+1, −64°, Ψi+1,
−27°, Φi+2, −90°, and Ψi+2, −7°; type VIII, Φi+1, −72°, Ψi+1,
−33°, Φi+2, −123°, and Ψi+2, 121°) indicates that the
conversion from type I to type VIII requires an ∼130° rotation
of the i+2 carbonyl group (Ψi+2), which links the L-arylGly
carbonyl group to the 1,4-benzamide of the following turn. This
analysis suggests that the energy difference between type I and
type VIII is sufficiently small as to be compensated by the
CH−π interactions between the Pro and arylGly residues and
two inter-ring hydrogen bonds (β-4(N)→β-4′(O) and β-
4′(N)→β-4(O), Figure 6).
H
Stereoisomeric L-Pro-L-arylGly (aryl = naphthyl, 2; phenyl,
H2Ph) dipeptides, i.e., L-Pro-D-arylGly, D-Pro-D-arylGly, and D-
Pro-L-arylGly dipeptides, were investigated to probe the validity
of the above suggestion. In combination with D-Pro-Aib (D-1a),
each of these cases generated only non-catenated cyclic
macrocycles. Since the i+1 and i+2 configurations of the L,D,
D,D, and D,L dipeptides favor type II, I′, and II′ β-turns,
respectively,13c,23,30c their metrical biases32 would require
significant conformational change to both residues (Pro and
arylGly) to achieve the optimal β-4 structure, differences which
might not be compensated for through weak secondary forces.
3.2. L-Pro Modification in the L-Pro-L-arylGly unit. As
depicted in Figure 1 and Scheme 2a, the structure of H3a
supports a CH−π interaction between a proline unit and the
naphthyl group (β-4 to β-4′). Since the acidity and geometric
arrangement of these CH donors should perturb the CH−π
interaction,33 we investigated the effect of electronegative
11437
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Figure 6. X-ray structure of H3a displaying the inter-ring hydrogen
bonds between the β-4 and β-4′ backbones (red dotted lines). The red
circles indicate the rotation (the black arrow) of the L-NaphGly
carbonyl group (i+2 residue) in the β-4 turn for the turn conversion
(type I to type VIII). All hydrogen atoms except those forming inter-
ring hydrogen bonds are omitted for clarity.
Figure 7. Generation of [2]-catenanes from D-1a and L,L-y2 or L,L-y2Ph.
aOther constitutional isomers of 3x and 4x. bAlthough precipitation of
noncatenated products occur after 3−5 days, the catenane reaches a
steady state within 3 days (see the details in SI1 Figures 13 and 14).
substituents at H1 and H2 (Scheme 2a). Supporting this
hypothesis are experimental studies by Zondlo showing that
inductive effects enhance proline CH−π interactions,12a and
computational studies by noting the benefit of α-halo
substituents on CH−π binding strengths.33b
ring pucker has been known to bias the backbone torsional
angles of adjacent residues,38 it cannot be excluded that cis-4-
substituted prolines with the Cγ-endo arrangement prefer a β-
turn that is not type I, thus causing a destabilization of the
catenane (vide supra).39
To this end we examined DSA reactions with mixtures of D-
Pro-Aib (D-1a) and variants of L-yPro-L-NaphGly (y = OH, F),
where H1 is either OH or F (trans-y-L-proline; (2S,4R)-4-y-L-
proline) or where H2 is OH (cis-hydroxy-L-proline; (2S,4S)-4-
hydroxy-L-proline) (Scheme 2a), or similarly, using the L-yPro-
L-PhGly analogues.34 In all cases, DSA reactions with polar H2
substituents (i.e., cis-OH or cis-F) provided no detectable
catenane. However, in the H1 replacement series (trans-OH
and trans-F), significant quantities of the [2]-catenanes were
generated (Figure 7).35 Complicating this analysis was the
onset of precipitation after 3−5 days for these monomers,
though the kinetic evolution of catenane formation clearly
shows that [2]-catenane concentrations reach a steady state
before the onset of precipitation (see SI1 Figures 13 and 14).
Since electronegative groups at H1 (trans-y) increase the
positive charge on the interacting hydrogens, this substitution
should intensify the electrostatic component of the CH−π
interaction.33a,36 An additional factor that may play a minor role
is the fact that trans-4-substituted prolines favor the Cγ-exo ring
pucker, which is the observed pucker in all structurally
characterized catenanes.3,37 The growth of [2]-catenane
concentration on introducing electronegative elements at H1
implies a slightly stronger CH−π interaction. The perturbation
is especially noticeable when the nucleating aromatic is phenyl.
The lack of catenane with electronegative H2 substituents (cis-
y) likely reflects a destabilizing electrostatic repulsion between
the naphthyl π system and the proximal OH or F atom. The Cγ-
endo arrangement of cis-4-substituted prolines37 likely results in
an additional destabilization both by depopulating the more
favorable pucker, and by also positioning the electronegative
group closer to the aryl π−donor. However, since the proline
Unexpectedly, the constitutional selectivity of the catenane is
sensitive to the nature of the H1 substituent, with trans-OH
decreasing the amount of 4 and trans-F increasing it. In the
former case, HO3a accounts for more than 95% of the total
catenanes. The higher propensity for forming the non-C2
product 4 in the trans-F case was consistently observed in
DSA reactions of D-monomers and L-trans‑FPro-L-NaphGly
(L,L-trans‑F2, Figure 8). In all paired DSA reactions (Figure 8),
the equilibrium concentration of the non-symmetric [2]-
catenane, [F4x]eq, was higher than [H4x]eq; the total yield of
F
F[2]-catenanes was also higher. In particular, 4x accounted for
∼50% of all catenanes in DSA reactions generated from
cycloaliphatic monomers (D-1r−D-1u, Scheme 3). The intrinsic
preference for H4x in these monomers (Table 2, entries 11−
14) and F4x favoritism in fluorine-substituted monomers is
additive, as judged by the high [2]-catenane concentrations for
1s, 1t, and 1u in Figure 8. This effect was utilized to advantage
3
F
in isolating a pure sample of 4u for X-ray analysis.
The X-ray structures in the previous paper3 indicate that the
proline−π interaction is created from two pseudo-axial CH’s
and one pseudo-equatorial CH interacting with the aryl ring.
Computational studies of CH−π interactions between cyclic
hydrocarbons and aromatics indicate that CH−π strengths
track with the number of contacting CH’s, although the total
strength is not precisely additive.40,41 To probe the effect of CH
donor number, the four-membered analogue of L-Pro was
examined. L-Azetidine-2-carboxylic acid (Aze) was employed
(L,L-H2Aze and L,L-H2PhAze) along with D- or D,L-1x (x = a, b, m,
u, v) to determine if reducing the number of CH donors would
decrease the amount of [2]-catenanes. This substitution is not
11438
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Table 3. [2]-Catenanes Generation from D- or D,L- (x = a, b,
m, u, v) and L,L-H2Aze
Figure 8. Generation of [2]-catenanes from D- or D,L-1a and L,L-H2
a
(dot-filled) or L,L-trans‑F2 (filled). Other constitutional isomers of 3x
and 4x.
a
Yields obtained from the UV trace of the corresponding catenane in
each DSA solution at day 7−9 using a calibration curve. The
calibration curve was obtained from the HPLC-UV area (289 nm) of
several known concentrations of pure catenanes; details in SI1 Figure
1. The values in the parentheses were obtained from 3:1 (D:L) biased
flawless, as Aze’s dihedral preferences and the projection of the
CH’s could be slightly different due to its flattened ring
pucker.42 In addition, a different turn preferences from L-Pro-X
cannot be excluded, as the turn preferences of the dipeptide L-
Aze-X are not known.
b
DSA solutions. The yield difference between total catenanes obtained
c
from DSAs employing L,L-H2 and L,L-H2Aze, respectively. The yields of
d
L,L-H2Aze and several monomers (Table 3) were combined
and the total steady-state concentrations determined. In general
the Aze monomer was tolerated, implying that L,L-H2Aze can
adopt the type VIII turn that L,L-H2 does. In the parent D-Pro-
Aib case, the decreased quantity of [2]-catenanes (from 72% to
47% of the total) implies a weakened CH−π interaction due to
either an imperfect arrangement or a reduction of the number
of CH donors. For entries 3 and 5, which position a 4-MeO-
C6H4-ring and a glycine on the opposite face of the naphthyl
ring, only traces of [2]-catenane were observed. When the
naphthyl group was replaced with phenyl (L,L-H2PhAze), all cases
gave <11% [2]-catenanes (SI1 Table 3). Thus, weakening the
π, reducing the number of proline CH donors, and removing
the optimal CH donors in the β-2 position (Scheme 2) causes
total or near total loss of [2]-catenane viability. For the
naphthyl cases, it appears that a minimum of three CH−π
interactions are needed for measurable stability; a Pro−π−Gly
case can be accommodated, but an Aze−π−Gly cannot.
Increasing the CH donors from the β-2 position in Aze−π−
Aib restores measurable stability, suggesting that the catenane
can compensate for the loss of some CH−π interactions, but
once a threshold of three has been reached, the structure is no
longer competitive. When the π is phenyl (2Ph series), the
threshold is higher.
other constitutional isomers of 3x and 4x were 3 (1)%. Precipitation
occurred within 3−5 days. The yields of the generated catenanes were
obtained at day 7−8 after filtering the precipitates.
DISCUSSION
■
In summary, mixtures of two dipeptides that are epimeric at
proline generate octameric [2]-catenanes under dynamic self-
assembly (DSA) conditions; the reactions are diastereoselective
and constitutionally selective. The solid-state and NMR
structural analysis of the [2]-catenanes discussed in the
previous paper3 reveal that the catenanes are stabilized by a
combination of intra- and inter-macrocyclic hydrogen bonds,
multiple aromatic interactions, and CH−π interactions. The
outcome of our structure/stability studies indicates that the
core structure is especially important, with the pair of CH−π−
CH sandwiches, the inter-macrocycle hydrogen bonds between
the β-2 and β-4 turns, and the predisposition for type II′ (for β-
2) and type VIII (for β-4) β-turns being important variables.
Indeed, the β-turn conformation at the core was found to be
the critical feature predicting stability, while the others
individually contribute to catenane stability; to various degrees
they could be weakened or removed with a concomitant cost.
Mutation studies in the core found that, when the proper
turns were accessible, the stability of the catenane relative to its
competing non-catenated structures could be reasonably
11439
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
described by a combination of non-covalent interactions.
Through methyl deletion experiments, proline to F-proline or
HO-proline mutations, and proline to azetidine carboxylic acid
mutations, it became apparent that a minimum of three CH−π
interactions (doubled by symmetry) are needed to compete
with non-catenated structures. In this regard, the CH−π
interactions in the pair of core CH−π−CH sandwiches
significantly contribute to product stability. Sufficiently
stabilizing CH−π interactions were additionally found to
compensate for the loss of the key cross-ring hydrogen bonds
when the D-Pro-L-Pro monomer was utilized (Figure 5). Each
of these highlighted examples supports the notion of an
interwoven network of energetically balanced non-covalent
interactions that can adjust their structure to compensate for
mutational weaknesses created in the structure.
Most illustrative of this concept is the X-ray analysis of the
catenane F3sPh (F3dPh in ref 3) obtained from the combination
of the D-Pro-Ac4c and L-FPro-L-PheGly monomers. A
comparative analysis of this structure and the parent H3a
indicates that conversion of the naphthyl to a phenyl ring in the
CH−π−CH sandwich causes the expected increase43 in the
CH−π distances due to a weakening of the π donor. In
response, the network of non-covalent interactions shortens the
adjacent inter-ring hydrogen bonds, which concomitantly
decreases the width of the aryl barrel and brings the β-1
loops sufficiently close that one of them flips from a type II′
turn to a type I′ to enable a new inter-ring hydrogen bond.
These compensating responses from the non-covalent network
help to recover some of the loss in stability that the naphthyl-
to-phenyl conversion would have otherwise created in the
CH−π−CH sandwich.
The cooperativity of the non-covalent network can be further
supported by a Gibbs free energy analysis. Since the number of
potential n-mers is the same for each DSA reaction, one can
semi-quantitatively ascribe an energy of the catenane relative to
the ensemble of competing structures by defining an
equilibrium between all competitors and the catenanes (eq
1). The resulting equilibrium constant (eq 2) can then be used
Figure 9. Diagram showing ΔΔG values (kcal/mol) between the six
DSAs. The values in parentheses are obtained from the CH−π
interaction values reported in the literature.12f In each DSA, ΔG was
obtained from the concentration ratio of catenanes and non-
catenanted species, and ΔΔG is the difference of two DSAs’ ΔG
values. See the details in SI1 Table 4.
the detection limit of the LC-UV traces. Thus, the D-Ala for Aib
substitution corresponds to the top right catenane and the
lowest energy structure (right green arrow in Figure 9). The
measured minimum energy difference is ∼5 kcal/mol. Since
each CH4---C10H8 interaction is worth 1.92 kcal/mol and two
are lost in the D-Ala for Aib mutation, a computational
prediction based on CH−π contributions only would predict a
ΔΔG of 3.84 kcal/mol. A similar analysis follows for the phenyl
series (left green arrow in Figure 9), except that two CH4---
C6H6 interactions are worth 2.90 kcal/mol, while the minimum
measured energy difference is ∼4.3 kcal/mol. The loss of
CH−π interactions is clearly insufficient to describe the
phenomenon. As discussed previously, this larger difference is
ascribed to the change in intrinsic β-turn preference for these
two dipeptides and the absolute requirement for a type II′ turn
at the β-2 position. Additionally, contributions no doubt come
from the loss of the aforementioned inter-ring hydrogen bonds
and the turn-on of the carbonyl−carbonyl repulsions at the i+1
position.
Keq
oligomers(cyclic, acyclic) HooI [2]‐catenanes
(1)
to provide a baseline energy, perturbations from which will
generate a ΔΔG that correlates to the perturbation gain or loss
in stability.44 A comparison of these perturbation free energy
changes to theoretical measures of the expected gain or loss of
stability provides an assessment of the independence of the
non-covalent interactions in question. Of particular interest are
theoretical measures of CH−π interaction strengths: the
computational interaction strength for the CH4---C6H6
complex is 1.45 kcal/mol, while it is 1.92 kcal/mol for the
CH4---C10H8 complex.12f
Results at the opposite end of the spectrum were also seen in
this data set. For example, when the Me1 and Me2 positions
(Scheme 2b) are each replaced with hydrogens, one should lose
the equivalent of two CH4---C10H8 interactions, i.e., 3.84 kcal/
mol of stability (recall the β-turn preferences are the same for
D-Pro-Aib and D-Pro-Gly). Instead, a ΔΔG of only 0.74 kcal/
mol is observed (right red arrow in Figure 9). If one assumes
the energy difference reflects a complete loss of the CH−π
interactions, this measured difference is most reasonably
interpreted as resulting from compensating forces. That is,
the network is sufficiently entwined that losses in one part of
[[2]‐catenanes]eq
[∑ non‐catenated oligomers]eq
Keq
=
(2)
Using the methyl deletion experiments outlined in Scheme
2b and reported in Table 1, one can predict how much of a
drop in stability to expect on changing a naphthyl to a phenyl,
or removing a CH3−π interaction. Shown in Figure 9 are the
relative energies of these same changes determined exper-
imentally using the approach in eqs 1 and 2. In the case of the
two D-Ala scenarios, these concentrations were estimates from
11440
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
the web can be neutralized by gains in another. Phenyl for
naphthyl changes were found to behave similarly (blue arrows
in Figure 9).
solutions, and calculations of the ΔΔG values between the six
DSAs using the oligomer speciation. This material is available
free of charge via the Internet at http://pubs.acs.org.
The complexity in the [2]-catenane leads to numerous
outcomes on mutagenesis. In some instances, the deletion of
seemingly important interactions (naphthyl to phenyl, Figure
9) causes little stability loss, as the network of stabilizing forces
can adjust the balance to compensate for the loss. In other
circumstances, the excision of a contributing interaction
(removal of CH3 in Scheme 2b) leads to disproportionate
losses in stability as larger primary forces are revealed. The
question of cooperativity and multi-valency is thus distorted by
the situation-dependent strength of the individual non-covalent
contributors to stability. That is, in a complex system, the
ground-state structure is a compromise (at the individual
interaction level) that maximizes the global balance of forces.
When the balance is shifted by a mutation, it is possible that a
new minimum can be achieved through a rearrangement that
best rebalances the energetic contributors.
The parallels to biological scenarios are many. For example, a
glutamine (Gln) for glutamic acid (Glu) mutation at the β-6
position of hemoglobin (Hb Machida)45 causes no conforma-
tional or functional changes. The replacement of the same
residue by lysine (Lys) or alanine (Ala) also leads to no
functional changes (Hb C and Hb G Makassar).46 By contrast,
replacement of this glutamic acid by valine (Hb S) causes a
severe structural change and a great diminishment of the
deoxygenated Hb S solubility. The subsequent polymerization
leads to sickle cell anemia.47 Thus, several mutations can be
accommodated, but one particular mutation leads to a shift
toward a different thermodynamic state with negative
consequences. The catenane mutagenesis recapitulates this
phenomenon.
AUTHOR INFORMATION
Corresponding Author
mgagne@unc.edu; mlwaters@unc.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the Defense Threat Reduction Agency (DTRA) for
support (HDTRA1-10-1-0030). S.J.L. thanks the Army
Research Office for support.
REFERENCES
■
(1) (a) Hunter, C. A.; Anderson, H. L. Angew. Chem., Int. Ed. 2009,
48, 7488−7499. (b) Whitty, A. Nat. Chem. Biol. 2008, 4, 435−439.
(c) Badjic, J. D.; Nelson, A.; Cantrill, S. J.; Turnbull, W. B.; Stoddart, J.
F. Acc. Chem. Res. 2005, 38, 723−732. (d) Williams, D. H.; Stephens,
E.; O’Brien, D. P.; Zhou, M. Angew. Chem., Int. Ed. 2004, 43, 6596−
6616. (e) Shinikai, S.; Ikeda, M.; Sugasaki, A.; Takeuchi, M. Acc. Chem.
Res. 2001, 34, 494−503. (f) Mammen, M.; Choi, S.-K.; Whitesides, G.
M. Angew. Chem., Int. Ed. 1998, 37, 2754−2794.
(2) Ercolani, G. J. Am. Chem. Soc. 2003, 125, 16097−16103.
(3) Chung, M.-K.; White, P. S.; Lee, S. J.; Waters, M. L.; Gagne, M.
R. J. Am. Chem. Soc. 2012, DOI: 10.1021/ja302345n.
(4) (a) Nishio, M. Phys. Chem. Chem. Phys. 2011, 13, 13873−13900
and references therein. (b) Takahashi, O.; Kohno, Y.; Nishio, M.
Chem. Rev. 2010, 110, 6049−6076 and references therein.
́
(5) (a) Carrillo, R.; Lopez-Rodríguez, M.; Martín, V. S.; Martín, T.
Angew. Chem., Int. Ed. 2009, 48, 7803−7809. (b) He, Q.; Rohani, S.;
Zhu, J.; Gomaa, H. Chirality 2012, 24, 119−128. (c) Saigo, K.;
Kobayashi, Y. Chem. Rec. 2007, 7, 47−56 and references therein.
(d) Suezawa, H.; Ishihara, S.; Umezawa, Y.; Tsuboyama, S.; Nishio, M.
Eur. J. Org. Chem. 2004, 23, 4816−4822. (e) Kobayashi, Y.; Kurasawa,
T.; Kinbara, K.; Saigo, K. J. Org. Chem. 2004, 69, 7436−7441.
(6) (a) Nishio, M.; Umezawa, Y.; Honda, K.; Tsuboyama, S.;
Suezawa, H. CrystEngComm 2009, 11, 1757−1788 and references
therein. (b) Schneider, H.-J. Angew. Chem., Int. Ed. 2009, 48, 3924−
3977 and references therein.
This [2]-catenane is an uncommon example of complex
folding phenomena in a biomimetic system. Rather than
isolating a single non-covalent interaction, as is the case in most
model systems, these catenanes provide a model for how
multivalent systems can compensate for unfavorable mutations
and minimize the “toxicity” of many,48 but not all mutations.
This is, of course, a crucial molecular-level characteristic that is
required for evolution.
(7) (a) Scarso, A.; Trembleau, L.; Rebek, J., Jr. J. Am. Chem. Soc.
2004, 126, 13512−13518. (b) Bhayana, B.; Wilcox, C. S. Angew.
Chem., Int. Ed. 2007, 46, 6833−6836. (c) For recognition involving
carbohydrates, see ref 4a and references therein.
(8) Nishio, M. Tetrahedron 2005, 61, 6923−6950 and references
therein.
CONCLUSION
■
Mutation studies of the key non-covalent interactions in the
[2]-catenanes described here suggest that the structures are
complex enough to mimic numerous key properties of large
protein systems. The manipulation of structure and concom-
itant changes in stability provide insight into the interplay
between the components of a network of non-covalent
interactions. Even in a self-assembled complex, the complexity
of the structure requires that a balance of forces be found, and
this balance can fluctuate as structural mutations occur. This
network of forces provides the means to compensate for non-
optimal changes or even explicitly reject these mutations that
create insurmountable energetic costs. These fundamental
properties are therefore not exclusively found in large biological
systems.
(9) Nishio, M. CrystEngComm 2004, 6, 130−158 and references
therein.
(10) (a) Ganguly, H. K.; Majumder, B.; Chattopadhyay, S.;
Chakrabarti, P.; Basu, G. J. Am. Chem. Soc. 2012, 134, 4661−4669.
(b) Wilger, D. J.; Park, J. H.; Hughes, R. M.; Cuellar, M. E.; Waters, M.
L. Angew. Chem., Int. Ed. 2011, 50, 12201−12204.
(11) Chung, M.-K.; White, P. S.; Lee, S. J.; Gagne,
Chem., Int. Ed. 2009, 48, 8683−8686.
́
M. R. Angew.
(12) (a) Thomas, K. M.; Naduthambi, D.; Zondlo, N. J. J. Am. Chem.
Soc. 2006, 128, 2216−2217. (b) Carroll, W. R.; Zhao, C.; Smith, M.
D.; Pellechia, P. J.; Shimizu, K. D. Org. Lett. 2011, 13, 4320−4323.
(c) Kim, E.-I.; Paliwal, S.; Wilcox, C. S. J. Am. Chem. Soc. 1998, 120,
11192−11193. (d) Fujii, A.; Hayashi, H.; Park, J. W.; Kazama, T.;
Mikami, N.; Tsuzuki, S. Phys. Chem. Chem. Phys. 2011, 13, 14131−
14141. (e) Nilsson Lill, S. O. J. Mol. Graphics Modell. 2010, 29, 178−
187. (f) Tsuzuki, S.; Fujii, A. Phys. Chem. Chem. Phys. 2008, 10, 2584−
2594. (g) Gil, A.; Branchadell, V.; Bertran, J.; Oliva, A. J. Phys. Chem. B
2007, 111, 9372−9379.
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthesis of monomers and their physical data, HPLC traces
including oligomer speciation, LC-MS, LC-TOF analyses of
various DSAs, yields of [2]-catenanes in the additional DSA
(13) (a) Das, C.; Naganagowda, G. A.; Karle, I. L.; Balaram, P.
Biopolymers 2001, 58, 335−346. (b) Raghothama, S. R.; Awasthi, S. K.;
11441
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
Balaram., P. J. Chem. Soc., Perkin Trans. 2 1998, 137−143. (c) Haque,
T. S.; Gellman, S. H. J. Am. Chem. Soc. 1997, 119, 2303−2304.
(d) Kantharaju; Raghothama, S.; Aravinda, S.; Shamala, N.; Balaram, P.
Biopolymers (Pept. Sci.) 2010, 94, 360−370. (e) Aravinda, S.; Harini, V.
V.; Shamala, N.; Das, C.; Balaram, P. Biochemistry 2004, 43, 1832−
1846.
Zanuy, D.; Haspel, N.; Tsai, C.-J.; Aleman
Biochemistry 2007, 46, 1205−1218.
́
, C.; Nussinov, R.
(28) Even when a weaker π-donor, L-Pro-L-PheGly, was employed,
significant amounts of [2]-catenanes were generated (SI1 Table 2).
(29) In a three-component DSA using D-Pro-D-Ala/D-Pro-Gly (1:1)
as the D-monomers with L,L-H2, the catenane compositions indicate the
complete interchangeability between D-Pro-Aib and D-Pro-Gly (see SI1
Figures 11 and 12).
(14) (a) Lai, J. R.; Huck, B. R.; Weisblum, B.; Gellman, S. H.
Biochemistry 2002, 41, 12835−12842. (b) Paul, P. K. C.; Sukumar, M.;
Bardi, R.; Piazzesi, A. M.; Valle, G.; Toniolo, C.; Balaram, P. J. Am.
Chem. Soc. 1986, 108, 6363−6370. (c) Harini, V. V.; Aravinda, S.; Rai,
R.; Shamala, N.; Balaram, P. Chem.Eur. J. 2005, 11, 3609−3620.
(d) Benedetti, E. Biopolymers (Pept. Sci.) 1996, 40, 3−44. (e) Toniolo,
C.; Crisma, M.; Formaggio, F.; Peggion, C. Biopolymers (Pept. Sci.)
2001, 60, 396−419.
(30) (a) Chatterjee, B.; Saha, I.; Raghothama, S.; Aravinda, S.; Rai,
R.; Shamala, N.; Balaram, P. Chem.Eur. J. 2008, 14, 6192−6204.
(b) Rai, R.; Raghothama, S.; Balaram, P. J. Am. Chem. Soc. 2006, 128,
2675−2681. (c) Nair, M.; Vijayan, M.; Venkatachalapathi, Y. V.;
Balaram, P. J. Chem. Soc., Chem. Commun. 1979, 1183−1184.
(31) (a) Park, H. S.; Kim, C.; Kang, Y. K. Biophys. Chem. 2003, 105,
89−104. (b) Antohi, O.; Sapse, A.-M. J. Mol. Struct. (Theochem) 1998,
430, 247−258. (c) Aubry, A.; Cung, M. T.; Marraud, M. J. Am. Chem.
Soc. 1985, 107, 7640−7647.
F
(15) 7−10 days for DSAs containing H2 or 2, 2−3 days for DSAs
using HO2 or HO2Ph, 3−4 days for DSAs having H2Ph or F2Ph, and 3−5
F
days for DSAs having H2Ph or 2Ph
.
Aze
Aze
(32) The standard dihedral angle comparison of these types and type
VIII shows more than 110° differences in Φi+1, Ψi+1, and Ψi+2 values,
which relate to the turn backbone amide and carbonyl (i+1) and
carbonyl (i+2) groups: type II, Φi+1, −60°, Ψi+1, 131°, Φi+2, 84°, and
(16) The standard dihedral angle comparisons of the type II′ and I′
turns show almost 180° differences in Ψi+1 and Φi+2 values, which
correspond to the turn backbone carbonyl (i+1) and amide (i+2)
groups: type II′, Φi+1, 60°, Ψi+1, −126°, Φi+2, −91°, and Ψi+2, 1°; type
I′, Φi+1, 55°, Ψi+1, 38°, Φi+2, 78°, and Ψi+2, 6°.
(17) Only a trace amount of [2]-catenanes (≪1%) is generated. The
loss was calculated from the HPLC-UV trace total area (289 nm)
comparison of [2]-catenanes generated from a DSA reaction using a D-
Pro-Aib (D-1a) and a L-Pro-L-NaphGly (L,L-H2).
(18) (a) McComas, C. C.; Crowley, B. M.; Boger, D. L. J. Am. Chem.
Soc. 2003, 125, 9314−9315. (b) Walsh, C. T.; Fisher, S. L.; Park, I.-S.;
Prahalad, M.; Wu, Z. Chem. Biol. 1996, 3, 21−28. (c) Walsh, C. T.
Science 1993, 261, 308−309. (d) Bugg, T. D. H.; Wright, G. D.; Dutka-
Malen, S.; Arthur, M.; Courvalin, P.; Walsh, C. T. Biochemistry 1991,
30, 10408−10415.
Ψi+2, 1°; type I′, Φi+1, 55°, Ψi+1, 38°, Φi+2, 78°, and Ψi+2, 6°; type II′,
Φi+1, 60°, Ψi+1, −126°, Φi+2, −91°, and Ψi+2, 1°.
(33) (a) Nishio, M.; Hirota, M.; Umezawa, Y. The CH/π Interaction:
Evidence, Nature, and Consequences; Wiley-VCH; New York, 1998.
(b) Fujii, A.; Shibasaki, K.; Kazama, T.; Itaya, R.; Mikami, N.; Tsuzuki,
S. Phys. Chem. Chem. Phys. 2008, 10, 2836−2843. (c) Gotch, A. J.;
Pribble, R. N.; Ensminger, F. A.; Zwier, T. S. Laser Chem. 1994, 13,
187−205. (d) Gord, J. R.; Garrett, A. W.; Bandy, R. E.; Zwier, T. S.
Chem. Phys. Lett. 1990, 171, 443−450.
(34) L,L-cis‑F2 and L,L-cis‑HO2Ph monomers could not be separated from
their epimers. Therefore, the experiments using these monomers were
omitted.
(19) Since the [2]-catenanes are D-component-rich, the residual
oligomers populate the more L-component-rich species with D-1a.
Despite this bias, the n-mer ratios are largely unaffected.
(20) An equivalent logical analysis can be constructed for the 4
catenanes.
(35) See SI2 for HPLC traces of DSA solutions and LC-TOF
analyses of the generated catenanes.
(36) (a) Dey, R. C.; Seal, P.; Chakrabarti, S. J. Phys. Chem. A 2009,
113, 10113−10118. (b) Tsuzuki, S.; Honda, K.; Uchimaru, T.;
Mikami, M.; Tanabe, K. J. Am. Chem. Soc. 2000, 122, 3746−3753.
(37) (a) Kim., W.; Hardcastle, K. I.; Conticello, V. P. Angew. Chem.,
(21) Since [2]-catenane 4 contains three units of L,L-H2, a separate
but logically equivalent set of isomers should result.
́
Int. Ed. 2006, 45, 8141−8145. (b) Lesarri, A.; Cocinero, E. J.; Lopez, J.
(22) Similarly consistent with case 2 was the detection (by LC-TOF)
of only two isomers of 3 with 1 equiv of D-Pro-D-Ala (Figure 3a; see
additional details in SI1 Figure 7).
C.; Alonso, J. L. J. Am. Chem. Soc. 2005, 127, 2572−2579. (c) Thomas,
K. M.; Naduthambi, D.; Tririya, G.; Zondlo, N. J. Org. Lett. 2005, 7,
2397−2400. (d) DeRider, M. L.; Wilkens, S. J.; Waddell, M. J.;
Bretscher, L. E.; Weinhold, F.; Raines, R. T.; Markley, J. K. J. Am.
Chem. Soc. 2002, 124, 2497−2505. (e) Renner, C.; Alefelder, S.; Bae, J.
H.; Budisa, N.; Huber, R.; Moroder, L. Angew. Chem., Int. Ed. 2001, 40,
923−925.
(23) Gibbs, A. C.; Bjorndahl, T. C.; Hodges, R. S.; Wishart, D. S. J.
Am. Chem. Soc. 2002, 124, 1203−1213.
(24) The introduction of a CF3 at Me 1 (Scheme 2b) (D-Pro-3,3,3-
trifluoro-L-Ala) completely destabilized the catenanes, and no catenane
formation was observed.
(38) (a) Okuyama, K.; Miyama, K.; Morimoto, T.; Masakiyo, K.;
(25) The Hunter−Sanders model predicts an increased π/π repulsion
with electron-donating substituents, indicating that electrostatic effects
alone do not explain the observed stability trend. X-ray analysis of
HO3n (HO3c in ref 3) did not reveal any close contacts to the para-
substituent. Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990,
112, 5525−5534.
Mizuno, K.; Bachinger, H. P. Biopolymers 2011, 95, 628−640.
̈
(b) Shoulders, M. D.; Satyshur, K. A.; Forest, K. T.; Raines, R. T.
Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 559−564. (c) Jiravanichanun,
N.; Nishino, N.; Okuyama, K. Biopolymers 2006, 81, 225−233.
(39) The fact that the cis-4-F-L-Pro-Gly segment has been reported to
more strongly favor the type I β-turn than trans-4-F-L-Pro-Gly may
attenuate this possibility: Kim, W.; McMillan, A.; Snyder, J. P.;
Conticello, V. P. J. Am. Chem. Soc. 2005, 127, 18121−18132.
(40) (a) Tsuzuki, S.; Honda, K.; Fujii, A.; Uchimaru, T.; Mikami, M.
Phys. Chem. Chem. Phys. 2008, 10, 2860−2865. (b) Morita, S.; Fujii,
A.; Mikami, N.; Tsuzuki, S. J. Phys. Chem. A 2006, 110, 10583−10590.
(c) Ringer, A. L.; Figgs, M. S.; Sinnokrot, M. O.; Sherrill, C. D. J. Phys.
Chem. A 2006, 110, 10822−10828.
F
(26) X-ray structure analysis of the catenane 3sPh (F3dPh in ref 3)
containing the β-turns of Ac4c residues also shows that these β-turns
are quite similar to those of Aib. Previous studies have shown Ac3c
residues to form the similar (but more rigid) β-turns: (a) Aleman, C. J.
́
Phys. Chem. B 1997, 101, 5046−5050. (b) Gatos, M.; Formaggio, F.;
Crisma, M.; Toniolo, C.; Bonora, G. M.; Benedetti, Z.; Di Blasio, B.;
Iacovino, R.; Santini, A.; Saviano, M.; Kamphuis, J. J. Pept. Sci. 1997, 3,
110−122. (c) Crisma, M.; Bonora, G. M.; Toniolo, C.; Barone, V.;
Benedetti, Z.; Di Blasio, B.; Pavone, V.; Pedone, C.; Santini, A.;
Fraternali, F.; Bavoso, A.; Lelj, F. Int. J. Biol. Macromol. 1989, 11, 345−
352. (d) Toniolo, C. Biopolymers 1989, 28, 247−257.
(41) Ran, J.; Wong, M. W. J. Phys. Chem. A 2006, 110, 9702−9709.
(42) (a) Blessing, R. H.; Smith, D. Acta Crystallogr. 1982, B38,
1203−1207. (b) Berman, H. M.; McGandy, E. L.; Burgner, J. W., II;
VanEtten, R. L. J. Am. Chem. Soc. 1969, 91, 6177−6182. (c) Jhon, J. S.;
Kang, Y. K. J. Phys. Chem. B 2007, 111, 3496−3507.
(43) Computational studies show that the optimum aromatic−CH
intermolecular separation (R) increase from the pyrene−methane (R =
3.4 Å) to the naphthalene−methane (R = 3.6 Å) to the benzene−
(27) (a) Ballano, G.; Zanuy, D.; Jimen
Nussinov, R.; Aleman, C. J. Phys. Chem. B 2008, 112, 13101−13115.
(b) Zanuy, D.; Rodrígues-Ropero, F.; Haspel, N.; Zheng, J.; Nussinov,
R.; Aleman, C. Biomacromolecules 2007, 8, 335−3146. (c) Zheng, J.;
́
ez, A. I.; Cativiela, C.;
́
́
11442
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443

Journal of the American Chemical Society
Article
methane (R = 3.8 Å) clusters, respectively. R corresponds to the
distance between the aromatic plane and the methane carbon atom.
See refs 40a and 12f.
(44) In practice this was accomplished by carrying out a full
speciation analysis and expressing the catenane concentration over the
sum of the other oligomer concentrations (see SI1 Table 4).
(45) Harano, T.; Harano, K.; Ueda, S.; Shibata, S.; Imai, K.; Seki, M.
Hemoglobin 1982, 6, 531−535.
(46) (a) Hunt, J. A.; Ingram, V. M. Biochim. Biophys. Acta 1960, 12,
409−421. (b) Blackwell, R. Q.; Oemijati, S.; Pribadi, W.; Weng, M.-I.;
Liu, C.-S. Biochim. Biophys. Acta 1970, 214, 396−401.
(47) (a) Ingram, V. M. Nature 1957, 180, 326−328. (b) Ingram, V.
M. Biochim. Biophys. Acta 1959, 36, 402−411. (c) Bihoreau, M. T.;
Baudin, V.; Marden, M.; Lacaze, N.; Bohn, B.; Kister, J.; Schaad, O.;
Dumoulin, A.; Edelstein, S. J.; Poyart, C.; Pagnier, J. Protein Sci. 1992,
1, 145−150. (d) Baudin-Chich, V.; Pagnier, J.; Marden, M.; Bohn, B.;
Lacaze, N.; Kister, J.; Schaad, O.; Edelstein, S. J.; Poyart, C. Proc. Natl.
Acad. Sci. U.S.A. 1990, 87, 1845−1849. (e) Adachi, K.; Kim, J.; Travitz,
R.; Harano, T.; Asakura, T. J. Biol. Chem. 1987, 262, 12920−12925.
(f) Padlan, E. A.; Love, W. E. J. Biol. Chem. 1985, 260, 8272−8279.
(g) Magdoff-Fairchild, B.; Poillon, W. N.; Li, T.-I.; Bertles, J. F. Proc.
Natl. Acad. Sci. U.S.A. 1976, 73, 990−994. (h) Harrington, D. J.;
Adachi, K.; Royer, W. E., Jr. J. Mol. Biol. 1997, 272, 398−407.
(i) Nagel, R. L.; Johnson, J.; Bookchin, R. M.; Garel, M. C.; Rosa, J.;
Schiliro, G.; Wajcman, H.; Labie, D.; Moo-Penn, W.; Castro, O.
Nature 1980, 283, 832−834.
(48) The only two fully “toxic” mutations are the change in the β-
turn conformation at β-2 and β-4 and the incorporation of a polar
group in a position that will interact with the π-system. Other typical
types of mutations, such as changes in steric and electronic interactions
at the Aib position, “knock down” the formation of the catenane, but
do not completely knock it out.
11443
dx.doi.org/10.1021/ja302347q | J. Am. Chem. Soc. 2012, 134, 11430−11443