Peptide dimerization is ubiquitous in natural protein conjugates and artificial self-assemblies. A major challenge in artificial systems remains achieving quantitative peptide heterodimerization, critical for next-generation biomolecular purification and formulation of therapeutics. Here, we employ a synthetic host to simultaneously encapsulate an aromatic and a noncanonical l-perfluorophenylalanine-containing peptide through embedded polar-π interactions, constructing an unprecedented series of heteropeptide dimers. To demonstrate the utility, this heteropeptide dimerization strategy was applied toward on-resin recognition of N-terminal aromatic residues in peptides as well as insulin, both exhibiting high recycling efficiency (>95%). This research unveils a generic approach to exploit quantitative heteropeptide dimers for the design of supramolecular (bio)systems.
Peptide dimerization is ubiquitous in natural protein conjugates and artificial self-assemblies. A major challenge in artificial systems remains achieving quantitative peptide heterodimerization, critical for next-generation biomolecular purification and formulation of therapeutics. Here, we employ a synthetic host to simultaneously encapsulate an aromatic and a noncanonical l-perfluorophenylalanine-containing peptide through embedded polar-π interactions, constructing an unprecedented series of heteropeptide dimers. To demonstrate the utility, this heteropeptide dimerization strategy was applied toward on-resin recognition of N-terminal aromatic residues in peptides as well as insulin, both exhibiting high recycling efficiency (>95%). This research unveils a generic approach to exploit quantitative heteropeptide dimers for the design of supramolecular (bio)systems.
Peptide dimerization through either covalent or noncovalent
bonding is key in structural design and functional control of natural[1] and artificial[2] self-assembly.
Covalent conjugation requires elaborate reactions to form static,
strong covalent bonds between peptides,[3] while noncovalent peptide dimerization is facile and versatile,[4] on account of its dynamic and reversible nature.
However, the relatively weak association and low specificity have
limited its use in aqueous systems. To address this problem, various
synthetic hosts have been utilized to encapsulate hydrophobic peptide
residues within their nanocavities, enhancing the overall binding
strength of noncovalent peptide dimers.[5]On account of their high binding affinity and range of guests,
cucurbit[n]uril (CB[n]) macrocyclic
hosts are ideal to bind peptides.[6−11] Urbach and co-workers reported a homopeptide dimer between two FGG
tripeptides and CB[8], displaying high binding strength (K ≈ 1011 M–2).[6] This homodimer has been adopted as a versatile building
block in the design and fabrication of supramolecular oligomers,[12] polymers,[8]uril-based
highly specific host–guest interactions: An approach to the
assembly of functional proteins. Angew. Chem.,
Int. Ed.. 2013 ">13,8]uril-mediated supramolecular
crosslinking of branched polyethylenimine. Chem.
Commun.. 2017 ">14] hydrogels,[15,16] and protein/peptide assemblies.[8]uril and dimeric adapter protein
14–3-3. Angew. Chem., Int. Ed.. 2017 ">17] Although
significant advances have been made, a major challenge remains favorable,
quantitative formation of heteropeptide dimers without homodimerization.Herein, we employ CB[8] to mediate heterodimerization of a canonical
aromatic peptide and a noncanonical l-perfluorophenylalanine(F′)-containing
peptide, Figure a.
Recently, we reported that an electron-poor perfluorophenyl first
guest and an electron-rich phenyl second guest can exclusively form
a CB[8]-mediated heteroternary complex through host-enhanced polar−π
interactions.[18,19] Thus, we postulated that the
F′-containing peptide (F′GG) would exclusively form
a 1:1 complex with CB[8], avoiding homodimerization on account of
the electrostatic repulsion within a 2:1 complex. Subsequent association
of various aromatic peptides (e.g., WGG) with the F′GG-CB[8]
complex may enable access to a new host-enhanced heteropeptide dimer
with superior binding strength, Figure a.
Figure 1
Schematic of (a) host-enhanced heteropeptide dimer formed
from
F′GG, WGG, and CB[8]; (b) molecular structures of FGG, YGG,
NpGG; (c) on-resin recognition through interfacial heterodimerization.
Schematic of (a) host-enhanced heteropeptide dimer formed
from
F′GG, WGG, and CB[8]; (b) molecular structures of FGG, YGG,
NpGG; (c) on-resin recognition through interfacial heterodimerization.Five model aromatic tripeptides containing l-perfluorophenylalanine[20] (F′GG), L-tryptophan (WGG), l-phenylalanine (FGG), l-tyrosine (YGG), and l-(2-naphthyl)alanine (NpGG) at the
N-termini were designed and prepared, Figure a–b. An equimolar mixture of F′GG, CB[8], and XGG
should result in an exclusive heterodimer instead of an equilibrium
mixture containing homodimers. Two additional series of tripeptides
(GGX, GXG), containing aromatic amino acids
either at the C-termini or in the midchain, were synthesized to investigate
a range of second guests and the effect of their position in the oligopeptides, Chart S1. After elucidating binding thermodynamics,
we applied this heterodimerization to achieve on-resin recognition and isolation[21,22] of aromatic tripeptides
from a peptide mixture exhibiting high efficiency and selectivity, Figure c.1H and 19F NMR titrations were performed
to probe heteropeptide dimerization within CB[8], Figure a and Figures S1–S16. Titration of WGG into a 1:1 mixture of F′GG-CB[8]
resulted in a gradual appearance of indole protons at 6.25–7.10
ppm. On account of shielding from the CB[8] cavity, these proton peaks
exhibited upfield shifts compared to free WGG, suggesting that the
indole group of F′GG-CB[8]-WGG is located in a different chemical
environment from unbound WGG. This titration was also monitored by 19F NMR, Figure a; a new group of fluorine peaks gradually appeared, while the peaks
of F′GG-CB[8] disappeared. Additionally, equivalent mixtures
of F′GG, CB[8], and XGG were characterized by high-resolution
ESI-MS, Figure b.
All ion peaks for F′GG-CB[8]-WGG, F′GG-CB[8]-FGG, F′GG-CB[8]-YGG, and F′GG-CB[8]-NpGG complexes were identified at their calculated m/z values. Together, these data confirmed the successful
formation of new heteropeptide dimers.
Figure 2
(a) 1H and 19F NMR spectra (D2O, 298 K) obtained through titration
of WGG (16.0 mM) into F′GG-CB[8]
(1.0 mM); (b) HR ESI-MS spectra of heteropeptide dimers (H2O, 1.0 mM) of WGG, FGG, YGG, NpGG with 1:1 complex of F′GG-CB[8];
ITC titration plots (10 mM phosphate buffer, pH 7.0, 298 K) of (c)
F′GG (3.0 mM) into CB[8] (0.1 mM); (d) WGG, FGG, YGG, NpGG
(3.0 mM) into F′GG-CB[8] (0.2 mM); and (e) schematic of self-sorting
mechanism.
(a) 1H and 19F NMR spectra (D2O, 298 K) obtained through titration
of WGG (16.0 mM) into F′GG-CB[8]
(1.0 mM); (b) HR ESI-MS spectra of heteropeptide dimers (H2O, 1.0 mM) of WGG, FGG, YGG, NpGG with 1:1 complex of F′GG-CB[8];
ITC titration plots (10 mM phosphate buffer, pH 7.0, 298 K) of (c)
F′GG (3.0 mM) into CB[8] (0.1 mM); (d) WGG, FGG, YGG, NpGG
(3.0 mM) into F′GG-CB[8] (0.2 mM); and (e) schematic of self-sorting
mechanism.Isothermal titration calorimetry
(ITC) was employed to study binding
thermodynamics of heteropeptide dimerization (Figure c and Figure S21–S25). Titration of F′GG (3.0 mM) into CB[8] (0.1 mM) led to a
stepwise binding curve with two transitions at molar ratios of 1.0
and 2.0 (Figure c).
The first-step binding constant (K1) is
6.6 × 105 M–1, while the second-step
(K2) is 9.4 × 103 M–1. This indicates negative cooperativity where K1 > K2.[18,19,23] Such favorable 1:1 complexation
enables secondary access of electron-rich aromatic peptides. Titrations
of XGG (X = W, F, Y, Np; 3.0 mM) into F′GG-CB[8] (0.2
mM) resulted in four binding curves with a clear transition at 1.0
molar ratio (Figure d), indicating successful incorporation of XGG into F′GG-CB[8].Table shows that
all XGG peptides exhibited high binding strengths (Ka > 104 M–1), confirming
thermodynamic stability of heteropeptide dimerization. YGG (Figure d, green) displayed
a relatively low Ka, as the para-substituted
hydroxyl group may decrease enthalpic contributions, weakening the
second association.[24] Nevertheless, the
overall binding constants (K1·Ka) for the heteropeptide dimers F′GG-CB[8]-XGG are all higher than 1010 M–2. This shows significant enhancement
compared to their parent dimers (e.g., F′GG-WGG,[25,26]Kdimer ≈ 1 M–1). The overall K1·Ka for F′GG-CB[8]-FGG (heterodimer 2.4 × 1011 M–2) is
higher than that for 2FGG-CB[8] (homodimer, 1.5 × 1011 M–2),[6] on account of
the enhanced polar−π interactions. Notably, no secondary
association of nonaromatic analogs (KGG, EGG, LGG) was observed, highlighting
selectivity for aromatic over nonaromatic peptides.
Table 1
Thermodynamic Data for Secondary Association
of Aromatic Residue-Containing Tripeptides and Control Tripeptides
with the 1:1 Complex of F′GG-CB[8]a
model peptide
Ka (103 M–1)
ΔHa (kcal mol–1)
–TΔSa (kcal mol–1)
WGG
460 ± 53
–13.3 ± 0.2
5.6 ± 0.2
GWG
103 ± 6
–11.9 ± 0.1
5.0 ± 0.1
GGW
90 ± 7
–14.5 ± 0.3
7.8 ± 0.4
FGG
356 ± 26
–14.4 ± 0.4
6.8 ± 0.3
GFG
58 ± 3
–9.1 ± 0.4
2.6 ± 0.4
GGF
22 ± 1
–7.5 ± 0.2
1.5 ± 0.2
YGG
3 ± 5
–9.6 ± 0.1
3.6 ± 0.2
GYG
4 ± 0.3
–4.7 ± 0.2
–0.2 ± 0.3
GGY
–b
–b
–b
NpGG
382 ± 22
–14.6 ± 0.2
7.0 ± 0.2
GNpG
165 ± 10
–12.7 ± 0.4
5.6 ± 0.4
GGNp
90 ± 2
–12.0 ± 0.1
5.2 ± 0.1
KGG
–b
–b
–b
EGG
–b
–b
–b
LGG
–b
–b
–b
Averaged with three replicates.
Not detected.
Averaged with three replicates.Not detected.To understand the influence of aromatic position on
heterodimerization, Ka values of GXG and
GGX (X = W, F, Y, Np) with
F′GG-CB[8] were determined by ITC. Shifting aromatic residues
from N- to C-termini led to a notable decrease in Ka, Table . The increased distance between the positive charge at the N-terminus
and the aromatic motif weakens ion–dipole interactions at the
CB[8] portal, reducing the secondary binding affinity. This is exemplified
by GGY, where no secondary binding to F′GG-CB[8] was observed.
Seven new, derivatized heteropeptide dimers, F′GG-CB[8]-GXG and F′GG-CB[8]-GGX, expand the scope of host-enhanced heteropeptide
dimerization. The exhibited binding selectivity to aromatic residues
is an advantage of this system, enabling access to a range of peptides
and proteins. Compared to previous reports on CB[8]-peptide heteroternary
complexes,[27] the system described here
is simply based upon an F′ amino acid, easily accessible for
ligation in chemical biology and biochemistry.The thermodynamic
mechanism behind peptide heterodimerization is
attributed to social self-sorting, Figure e, consistent with previous reports.[18,19] Two pathways exist in an equimolar mixture of F′GG, CB[8],
and XGG: social self-sorting (heterodimerization) and narcissistic
self-sorting (homodimerization). Compared to homodimerization (Table S1), social self-sorting was favored in
the presence of a ratio of hetero- and homopeptide dimers (e.g., WGG,
R = 2291), confirming quantitative heteropeptide dimerization in the
complex mixture. Notably, simply mixing two aromatic tripeptides with
CB[8] does not lead to quantitative heterodimerization (Figures S17–S19).After elucidation
of thermodynamics, we demonstrated the utility
of heteropeptide dimers to achieve on-resin recognition of aromatic
peptides, Figure c.
Confocal fluorescence imaging was employed to probe interfacial recognition
through a fluorescent tagged peptide (WGGGGG-dansyl),[9]Figure a–b. A buffered solution of WGGGGG-dansyl-CB[8] (10 mM) was
mixed with F′GGGGG-functionalized ChemMatrix resin (35–100 mesh particle size, surface loading
= 0.5 mmol/g) for 10 min at 25 °C with vigorous shaking. Confocal
fluorescent images were captured under a gray field upon laser irradiation
(λ = 405 nm). Figure c–f show that the resin with F′GGGGG-CB[8]-WGGGGG-dansyl
displays significantly higher fluorescence
than F′GGGGG, F′GGGGG-CB[8]-WGG, and F′GGGGG-WGGGGG-dansyl (Tables S3–S4), indicating the successful heteropeptide dimerization on resin.
Figure 3
Schematic
of (a) molecular structure of WGGGGG-dansyl, and (b)
interfacial recognition with F′GGGGG resin (10 mM); confocal
fluorescent images of (c) F′GGGGG resin, and those added with
(d) WGG-CB[8] (10 mM), (e) WGGGGG-dansyl (10 mM), and (f) WGGGGG-dansyl-CB[8]
(10 mM) obtained through laser excitation at λ = 405 nm under
a gray field. Scale bar = 200 μm. Fluorescence intensity is
quantified by mean gray value (MGV).
Schematic
of (a) molecular structure of WGGGGG-dansyl, and (b)
interfacial recognition with F′GGGGG resin (10 mM); confocal
fluorescent images of (c) F′GGGGG resin, and those added with
(d) WGG-CB[8] (10 mM), (e) WGGGGG-dansyl (10 mM), and (f) WGGGGG-dansyl-CB[8]
(10 mM) obtained through laser excitation at λ = 405 nm under
a gray field. Scale bar = 200 μm. Fluorescence intensity is
quantified by mean gray value (MGV).UV experiments were performed to test absorption efficiency through
quantification of aromatic peptides present before and after on-resin
treatment, Figure a. A typical experiment involved mixing WGG-CB[8] (1.0 mM) with F′GGGGG-resin
(10.0 mM) at 25 °C for 10 min. The absorption intensity of the
resin-treated WGG-CB[8] (gray) showed a decrease compared to the original
(purple), Figure b. The absorption efficiency for on-resin
recognition was 77%, while recognition of free WGG by physical absorption
was only 19%, Table S5. Absorbed WGG was
released and recycled through competitive binding by memantine hydrochloride
(DMADA), Figure S29, regenerating the resin.
Figure 4
(a) Schematic
of on-resin recognition and separation of an aromatic
tripeptide from a peptide mixture; (b) UV spectra of WGG-CB[8] (1.0
mM) before and after treatment with F′GGGGG-resin (10.0 mM);
histograms of (c) recycling efficiency of WGG in continuous on-resin
recognition cycles; (d) percentage of WGG obtained after multicycle
isolation; (e) schematic of on-resin stabilization of insulin, on-demand
release and binding to CB[8], not drawn to scale; (f) UV spectra of
insulin (0.2 mM) before and after treatment with F′GGGGG-CB[8]-resin
(10.0 mM) followed by release at days 0, 2, 5, and 8; (g) histograms
of recycling efficiency of insulin at days 0, 2, 5, and 8.
(a) Schematic
of on-resin recognition and separation of an aromatic
tripeptide from a peptide mixture; (b) UV spectra of WGG-CB[8] (1.0
mM) before and after treatment with F′GGGGG-resin (10.0 mM);
histograms of (c) recycling efficiency of WGG in continuous on-resin
recognition cycles; (d) percentage of WGG obtained after multicycle
isolation; (e) schematic of on-resin stabilization of insulin, on-demand
release and binding to CB[8], not drawn to scale; (f) UV spectra of
insulin (0.2 mM) before and after treatment with F′GGGGG-CB[8]-resin
(10.0 mM) followed by release at days 0, 2, 5, and 8; (g) histograms
of recycling efficiency of insulin at days 0, 2, 5, and 8.Multicycle on-resin recognition was performed to evaluate
recyclability, Figure c and Table S6. Recognition-regeneration
experiments
on WGG-CB[8] were repeated for 3 cycles using the same batch of resin.
The on-resin recycling efficiency was maintained above 98% over multiple
cycles (Figure c),
on account of complete release of WGG without any residue accumulation.
This confirms regeneration of F′GGGGG-functionalized resin,
endowing the whole process with high sustainability for practical
use. We further investigated selective isolation of aromatic peptides
through recognition-release experiments over 3 cycles (Figure d and Table S7), using a peptide mixture of WGG, KGG, EGG, and LGG ([XGG]
= 1.0 mM) in the presence of 1.0 mM CB[8]. The ratio of WGG (purple)
within the mixture is increased from 26 to >95% after 3 cycles
(Figure d). Additionally,
residual DMADA and DMADA-CB[8] can be removed through liquid-phase
chromatography. This facile strategy to obtain aromatic peptides with
high purity through on-resin heteropeptide dimerization is readily
amenable to automation.To extend applicability of this approach,
we exploited interfacial
recognition for insulin stabilization and its on-demand release from
the resin, Figure e and Table S8. Insulin is a widely used
biopharmaceutical for diabetes treatment;[28] however, on account of limited stability it requires strict storage
conditions (e.g., 2–6 °C) as it is prone to form immunogenic
fibrillar aggregates in solution.[29] Insulin
has an N-terminal phenylalanine, which can serve as a guest for CB[8].[30−32] Heteropeptide dimerization of insulin and F′GGGGG-functionalized
resin may offer a promising solution to address insulin instability.UV experiments quantified insulin absorbance onto the resin and
on-demand release. Insulin absorption efficiency was calculated to
be 94%, and its absorption intensity decreased after treatment with
F′GGGGG-CB[8] resin, Figure f and Figure S34. Through
competitive binding, insulin was displaced by DMADA with ∼95%
recycling efficiency (Figure g) over 8 days of storage at room temperature, indicating
long-term stability (Figure S35 and Table S9). Our approach provides a route for storing insulin under ambient
conditions, removing the current need for refrigeration.In
conclusion, we have introduced a new type of quantitative heteropeptide
dimerization. Through host-enhanced polar−π interactions,
the binding affinity between aryl and perfluorophenyl groups from
two different peptides is significantly enhanced with a Ka up to 105 M–1 and a K1·Ka up to
1011 M–2, ensuring exclusive formation
of heteropeptide dimers. To demonstrate utility, the solution-phase
host–guest complex (F′GG-CB[8]-XGG) was transferred
to a solid–liquid interface achieving on-resin recognition
and isolation of aromatic peptides as well as stabilization and on-demand
release of insulin under ambient conditions. This generic approach
enables accumulation and separation of aromatic-abundant biomacromolecules
useful in biomedical research. We anticipate that this work will inspire
research into exploitation of heteropeptide dimerization as a versatile
strategy for a wide range of life science applications.
Authors: Silvia Sonzini; Alessio Marcozzi; Raphael J Gubeli; Christopher F van der Walle; Peter Ravn; Andreas Herrmann; Oren A Scherman Journal: Angew Chem Int Ed Engl Date: 2016-10-13 Impact factor: 15.336
Authors: Matthew E Call; Jason R Schnell; Chenqi Xu; Regina A Lutz; James J Chou; Kai W Wucherpfennig Journal: Cell Date: 2006-10-20 Impact factor: 41.582
Authors: Leigh A Logsdon; Christopher L Schardon; Vijayakumar Ramalingam; Sharon K Kwee; Adam R Urbach Journal: J Am Chem Soc Date: 2011-10-03 Impact factor: 15.419
Authors: Jordan M Chinai; Alexander B Taylor; Lisa M Ryno; Nicholas D Hargreaves; Christopher A Morris; P John Hart; Adam R Urbach Journal: J Am Chem Soc Date: 2011-04-07 Impact factor: 15.419