Development and mechanistic studies of an optimized receptor for trimethyllysine using iterative redesign by dynamic combinatorial chemistry

Article abstract of DOI:10.1039/c4ob01249f

A new small molecule receptor, A₂N, has been identified that binds specifically to trimethyllysine (Kme₃) with sub-micromolar affinity. This receptor was discovered through the iterative redesign of a monomer known to incorporate through dynamic combinatorial chemistry (DCC) into a previously reported receptor for Kme₃, A₂B. In place of monomer B, the newly designed monomer N introduces an additional cation-π interaction into the binding pocket, resulting in more favorable binding to Kme₃ by 1.3 kcal mol^-1, amounting to a 10-fold improvement in affinity and a 5-fold improvement in selectivity over Kme₂. This receptor exhibits the tightest affinity and greatest selectivity for KMe ₃-containing peptides reported to date. Comparative studies of A ₂B and A₂N provide mechanistic insight into the driving force for both the higher affinity and higher selectivity of A₂N, indicating that the binding of KMe₃ to A₂N is both enthalpically and entropically more favorable. This work demonstrates the ability of iterative redesign coupled with DCC to develop novel selective receptors with the necessary affinity and selectivity required for biological applications. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4ob01249f

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
Development and mechanistic studies of an
optimized receptor for trimethyllysine using iterative
redesign by dynamic combinatorial chemistry†
Cite this: Org. Biomol. Chem., 2014,
12, 7059
Nicholas K. Pinkin and Marcey L. Waters*
A new small molecule receptor, A2N, has been identified that binds specifically to trimethyllysine (Kme3)
with sub-micromolar affinity. This receptor was discovered through the iterative redesign of a monomer
known to incorporate through dynamic combinatorial chemistry (DCC) into a previously reported receptor
for Kme3, A2B. In place of monomer B, the newly designed monomer N introduces an additional cation–π
interaction into the binding pocket, resulting in more favorable binding to Kme3 by 1.3 kcal mol−1, amount-
ing to a 10-fold improvement in affinity and a 5-fold improvement in selectivity over Kme2. This receptor
exhibits the tightest affinity and greatest selectivity for KMe3-containing peptides reported to date.
Comparative studies of A2B and A2N provide mechanistic insight into the driving force for both the higher
affinity and higher selectivity of A2N, indicating that the binding of KMe3 to A2N is both enthalpically and
entropically more favorable. This work demonstrates the ability of iterative redesign coupled with DCC
to develop novel selective receptors with the necessary affinity and selectivity required for biological
applications.
Received 16th June 2014,
Accepted 24th July 2014
DOI: 10.1039/c4ob01249f
www.rsc.org/obc
and hydrophobicity of the terminal ammonium group
increases while the hydrogen-bonding capacity, and thus the
Introduction
Tools that enable the identification and characterization of cost of desolvation, decreases. While these changes are subtle,
post-translational modifications (PTMs) in histone proteins reader proteins are capable of site-specifically recognizing
are essential for advancing the field of epigenetics. On the Kme, Kme2, or Kme3 and they facilitate downstream events in
histone tail, the site and extent of lysine (Lys) methylation is response to specific recognition events.8
linked to activation and repression of gene expression.1,2 More-
Antibodies are powerful and commonly employed tools for
over, dysregulation of Lys methylation has been associated characterizing PTMs. Due to their high specificity and well
with a number of types of cancers.3–7 Methylation can occur understood structure, they enable individual PTMs to be
up to three times, giving mono-, di- or trimethyllysine (Kme, tagged and enriched from complex samples, which is key to
Kme2, or Kme3, Fig. 1). With increasing methylation, the size methods such as ELISA, western blotting, and microarray
technology.9–12 Unfortunately, the high specificity is both a
strength and limitation; because site-specificity stems from
interactions with residues neighboring a specific PTM, it is
challenging to use antibodies to discover new sites of Lys
methylation. Furthermore, studies of the histone code are
often hampered by cross reactivity and interference with neigh-
boring PTMs.13–17
Synthetic small molecule receptors are emerging as promis-
ing new tools for studying PTMs.18–21 In the past decade, a
variety of interesting receptors have been reported that bind
Fig. 1 Methylation states of Lys that are found in proteins.
specifically to Kme3 through combinations of non-covalent
interactions. Several of these receptors bind tighter and more
Department of Chemistry, CB 3290, University of North Carolina at Chapel Hill,
selectively to Kme3 than native reader proteins, despite being
Chapel Hill, NC 27599, USA. E-mail: mlwaters@unc.edu
†Electronic supplementary information (ESI) available: Synthesis and characteri-
thousands of daltons smaller.22–27
We previously reported one such receptor, A2B (Fig. 2),
which was found to bind to Kme3 peptides in aqueous
zation of monomer N; NMR characterization of receptors; ITC binding data. See
DOI: 10.1039/c4ob01249f
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 7059–7067 | 7059

View Article Online
Paper
Organic & Biomolecular Chemistry
allows new monomers to be rapidly screened for their propen-
sity to incorporate into selective hosts, as the composition of
dynamic combinatorial libraries (DCLs) with varying guests is
often indicative of host selectivity. Using iterative monomer
redesign to optimize the affinity and selectivity of the lead
receptor, A2B, herein we describe the development and charac-
terization of a new receptor, A2N, with 300 nM affinity for
Kme3-containing peptides. Moreover, A2N exhibits ∼10-fold
tighter binding to Kme3, 5-fold greater selectivity over Kme2,
and >4-fold greater selectivity over unmethylated Lys relative to
A2B. The degree of affinity and selectivity of A2N makes it a
promising candidate to move forward with applications for
sensing Kme3. Moreover, analysis of the enthalpy and entropy
of binding to each of the methylation states of Lys to these two
receptors provides mechanistic insight into the factors provid-
ing affinity and selectivity.
Results and discussion
Fig. 2 Monomer B of A2B (top) was redesigned into monomer
N
System design
(bottom) to deepen the binding pocket of A2B and provide an extra
cation–π interaction.
A2B binds preferentially to Kme3 over the lower methylation
states of Lys via cation–π interactions in a binding pocket
made up of five aromatic rings.22 However, the selectivity over
Kme2 is a modest 2-fold. Computational modelling of A2B
suggested that the binding cavity is shallow, which may be
responsible for the low selectivity for Kme3 over Kme2 (Fig. 2).
We envisioned that a new monomer, N, if incorporated in
place of monomer B into a similar receptor, A2N, would
provide a deeper pocket and additional CH(δ+)–π interactions
with Kme3. Furthermore, we anticipated that Lys guests would
require greater desolvation to bind into the deeper binding
pocket of A2N, which we expected would improve selectivity for
Kme3.
solution with comparable affinity and selectivity to the HP1
chromodomain, a reader protein known to bind histone 3 (H3)
22,28
K9me2/3
.
A2B was identified using dynamic combinatorial
chemistry (DCC), a competitive selection method that simpli-
fies the discovery of synthetic receptors by relying on thermo-
dynamic equilibria to select for hosts that interact most
favorably with the guest of interest.29 Because the monomers
self-assemble into macrocyclic hosts, we proposed that we
could improve the binding and selectivity of A2B for Kme3 by
simply redesigning the constituent monomers. By focusing on
monomer redesign, the chemistry required to make changes to
a macrocycle’s binding pocket is simplified to the modification
of a small molecule. This reduces the challenges inherent in a
de novo approach to macrocycle modification, namely protect-
ing group optimization and targeted modification of single
functional groups when multiple identical groups exist. DCC
Synthesis
Monomer N was synthesized using an approach similar to that
reported by Otto and Sanders for synthesizing the isomeric
monomer A (Scheme 1).30 Initial efforts toward the dithio-
Scheme 1 Synthesis of monomer N.
7060 | Org. Biomol. Chem., 2014, 12, 7059–7067
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
carbamate anthracene 2 relied on previous reports of the syn-
thesis and modification of 1,4-anthracenediol, but were unsuc-
cessful due to rapid degradation of all intermediates.31
Instead, we found that the thiocarbamate group can act as a
protecting group for the reduction of the anthraquinone to the
anthracene, allowing the protected anthracene 2 to be reached
in acceptable yield over three steps. Compound 1 is first
reduced to the intermediate diol using NaBH4, then a reduc-
tive elimination using SnCl2 in aqueous acid and a subsequent
reprotection of the hydroxyl groups yields anthracene 2 (the
intermediate diol rapidly degrades in the presence of air and
light and is not isolated). The O-thiocarbamate anthracene 2 is
subjected to a Newman–Kwart rearrangement to yield the
S-thiocarbamate anthracene 3, which subsequently undergoes
a Diels Alder cycloaddition with dimethyl acetylene dicarboxy-
late (DMAD) to afford 4. A final base-promoted hydrolysis gives
monomer N cleanly and in high yield.
Library screening
DCC was used to rapidly screen for novel receptors for Kme3.
Disulfide exchange was used as the reversible reaction because
it occurs in aqueous solution at close to neutral pH and is
stable toward most biological functional groups.32 Dynamic
combinatorial libraries (DCLs) were set up with 2.5 mM each
of monomers A and N and guest concentration equal to the
total combined monomer concentration (i.e. 5 mM for a
2 monomer library) in 50 mM borate buffer, pH 8.5. Simple
peptides with the sequence Ac-KmeXGGL-NH2 (X = 0–3) were
used as guests to limit non-specific interactions that could
interfere with Lys recognition. Leu was incorporated to
decrease the polarity of the peptides, which simplified their
purification by reversed phase HPLC. For each combination of
monomers, five DCLs were set up in parallel: four with one of
the Lys guests and one untemplated library that lacked a
guest. DCLs were monitored by LC-MS after three days, twelve
days and three weeks. A species that was amplified in one
library more than any other library was pursued as a potential
selective receptor for the guest causing the amplification.
In DCLs containing only monomer A, A3 was amplified with
increasing methylation on Lys (Fig. S16†). In DCLs containing
only monomer N, no change in the library composition was
observed in the presence of any of the guests (Fig. S17†).
Fig. 3 (a) DCLs of monomers A and N with various guests (2.5 mM A,
2.5 mM N, 5 mM Ac-KmeXGGL-NH2, 50 mM borate buffer, pH 8.5)
(b) Calculated amplification of A2B and A2N by peak area in the presence
of different guests compared to the untemplated DCL.
Comparing the amplification of A2N and A2B (by peak area)
in similar DCLs, A2N is amplified 30-fold in the presence of
Kme3 over an untemplated library, while A2B is only amplified
about 10-fold (Fig. 3b). In the presence of the lower methyl-
ation states, similar amplification is observed for both recep-
tors. This observation suggests that A2N is a more selective
receptor for Kme3 than A2B.
Structural characterization
Instead, the monomer assembles into various forms of the The isomeric A2N macrocycles were made on a preparative
homocyclic tetramer N4. When both monomers are combined scale using DCC under similar conditions but with acetyl-
in equal concentrations, there is significant amplification of choline chloride (AcCh) as a template. AcCh was used instead
three species in the presence of Kme3 (Fig. 3a).
of Kme3 because it was found to similarly amplify A2N when
Because N is an isomer of A, it was impossible to identify present in excess in a DCL, yet is commercially available. After
by mass the exact identity of the three amplified species, all of five days, the isomers of A2N were isolated by RP-HPLC. Under
which were trimers. Nonetheless, a comparison to the DCLs of optimized conditions two isomers nearly co-elute (in 22%
the individual monomers (analyzed by LC-MS using the same yield), but the third isomer is better resolved and is easily iso-
method) suggested that the new species must be heterotrimers lated in 23% yield (Fig. S19†).
of A and N, since their retention times were different from A3,
Because monomer A is used in libraries as a racemic
and N3 was not amplified in the library of N (Fig. S18†). Once mixture, we expected that the three A2N species must be two
isolated, treatment of the receptors with TCEP resulted in distinct meso isomers and a pair of enantiomers. Initial experi-
reduction to a 2 : 1 mixture of monomers A and N, establishing ments revealed that at room temperature in methanol-d4 or
that the three species are all isomers of A2N (Fig. S30†).
D2O, the proton resonances of all three isomers of A2N were
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 7059–7067 | 7061

View Article Online
Paper
Organic & Biomolecular Chemistry
significantly broadened, indicating that all isomers of A2N are
dynamic and that rotation is on the NMR timescale. The para-
substitution of the thiols on N likely enables the monomer to
rotate about an axis created by the C–S bonds, making the
receptor quite flexible. While cooler temperatures only
increased the broadening, mild heating sharpened the reson-
ances significantly. In methanol-d4, less heating was required
to sharpen peaks compared to in D2O; therefore, all structural
characterization of A2N alone was performed in methanol-d4
(Fig. S22–S29†).
Table 1 Change in chemical shift (Δδ) observed for Ac-Kme3-Gly-NH2
upon binding to an excess of rac-A2B or meso2-A2N
Peptide protons
rac-A2B Δδ (ppm)
meso2-A2N Δδ (ppm)
Nme3
−1.59
−2.59
−2.11
−1.15
−0.58
−0.05
+0.18
+0.20
−2.46
−3.45
−3.25
−2.09
−0.60
+0.08
+0.25
+0.33
ε
δ
γ
β
A
Gly
Ac
A simple comparison of the 1H NMR spectra allowed the
two meso isomers to be assigned as the second and third
1
species that elute during purification, as their H spectra con-
further upfield. For both receptors, the ε protons exhibit the
greatest degree of upfield shifting and the extent of upfield
shifting of the other protons within the binding pocket
decreases with increasing distance from the ε protons. In con-
trast, there is no significant upfield shifting of any other
protons in the peptide, suggesting that the receptor interacts
primarily with the sidechain of Kme3.
tained fewer peaks than that of the first species. Because the
first meso isomer, meso1-A2N, co-elutes with rac-A2N, a pure
sample could not be obtained. However, the resonances of
meso1-A2N were distinguishable in the mixed spectrum, which
enabled further 2D NMR characterization. While rac-A2N could
be isolated with careful purification, there was significant peak
1
overlap in the H spectrum and further structural characteriz-
Comparing the upfield shifting of the methylene protons
between both receptors, the greatest difference in shift is
observed for the γ and δ methylenes, which meso2-A2N shifts
0.94 and 1.14 ppm further upfield than does rac-A2B, respect-
ively. Comparatively, the ε and Nme3 protons are both shifted
0.86 ppm further upfield by meso2-A2N. We expected that
incorporation of monomer N into A2N would result in a deeper
binding pocket that is capable of participating in additional
cation–π and CH(δ+)–π interactions with the methyl groups
and methylenes in the sidechain. This is evidenced by the
∼1 ppm further upfield shifting of all β, γ, δ, ε, and methyl
protons of Kme3 bound to meso2-A2N compared to rac-A2B.
ation was not pursued.
Proton assignments were made using the TOCSY and COSY
spectra of the two meso-isomers. The ROESY spectrum of
meso2-A2N revealed NOEs between protons 2, 3, & 4 on
monomer N and protons 10, 11, & 12 on monomer A, confirm-
ing the orientation of N in meso2-A2N as shown in Fig. 4. In
contrast, no inter-monomer NOEs were observed in the ROESY
spectrum of meso1-A2N. This suggests that meso1-A2N contains
a more open binding pocket than meso2-A2N, which may
help to explain the subsequent observation that meso2-A2N
binds tighter and more selectively than meso1-A2N to Kme3
(vide infra).
To determine the mode of binding to Kme3, an NMR analy-
sis of the dipeptide Ac-Kme3G-NH2 in the presence of excess
meso2-A2N was performed in D2O under saturating conditions.
Significant upfield shifting ranging from 0.6 to 3.5 ppm was
observed for the β, γ, δ, ε, and methyl protons of Kme3
(Table 1 and Fig. S32†), indicating close proximity of these
positions to the face of the aromatic rings of the receptor. This
is the largest degree of upfield shifting observed for any Kme3
receptor reported to date. Compared to rac-A2B, meso2-A2N
shifts the protons of Kme3 within its binding pocket ∼1 ppm
Binding studies
The previously reported Kd for rac-A2B binding to Kme3 in the
context of the histone 3 (H3) peptide, FAM-QTAR-K9me3-
STG-NH2 (where FAM is carboxyfluorescein), was determined
using fluorescence anisotropy (FA) to be 25 μM.22 To gain
more mechanistic insight into the driving force for binding,
we turned to isothermal titration calorimetry (ITC) to charac-
terize binding of A2N to Kme3. To make direct comparisons
between rac-A2B and A2N, we repeated measurements of the
binding of rac-A2B to the peptide H3 K9meX (Table 2), which
corresponds to residues 5–12 of histone 3, using ITC. The
binding to this sequence was studied for comparison to pre-
vious data for rac-A2B.22 A WGGG sequence was added to the
N-terminus of all peptides to enable concentration to be deter-
mined by UV. To verify that the receptors do not interact with
the Trp tag, binding was measured to an H3 peptide whose
Table 2 Peptides used for ITC titrations
H3 K9meX (X = 0–3)
H3 R8G-K9meX (X = 0–3)
H3 K9G
H3 R8G-K9G
H3 K36meX (X = 0, 3)
Ac-WGGG-QTARK
Ac-WGGG-QTAG
Ac-WGGG-QTARG
Ac-WGGG-QTAG
Ac-WGGG-TGGVK
̲
̲K̲
Fig. 4 Intra-monomer (left) and Inter-monomer (right) NOEs observed
for meso2-A2N. Numbering and NOEs are identical for each half of the
σ symmetric receptor.
̲
̲G̲
̲
7062 | Org. Biomol. Chem., 2014, 12, 7059–7067
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 3 Thermodynamic data obtained for the binding of rac-A2B and meso2-A2N to the peptides shown in Tables 2 as measured by ITCa
a Conditions: 26 °C in 10 mM borate buffer, pH 8.5. b Errors are from averages. c Selectivity is calculated as the factor-fold difference in affinity for
Kme3 over the designated methylation state in that row. d These values are approximate because the c-value for these experiments was <1. e For
these experiments, the N-value was fixed at 1 for one-site fitting. f NB = no binding.
basic Arg8 and Lys9 residues were mutated to Gly in order to
Inspection of ΔH and ΔS indicate that the difference in
eliminate any cation–π or charge–charge interactions with the affinity is due to small improvements in both the enthalpy and
receptors (Tables 2 and 3, entries 10 and 18). These control entropy of binding for meso2-A2N relative to rac-A2B. This goes
experiments verified that the Trp tag does not bind to rac-A2B against the typical trend of enthalpy–entropy compen-
or A2N.
sation.39,40 The more favorable enthalpy of meso2-A2N binding
Interestingly, ITC measurements of the binding of rac-A2B is most easily explained by greater van der Waals and cation–π
to H3 K9meX gave affinities to all methylation states that are interactions with the Kme3 sidechain. The more favorable
∼10-fold tighter than previously reported by FA (Table 3, entropy observed for meso2-A2N may be due to a greater contri-
entries 11–14), with a Kd of 2.6 μM for H3 K9me3, although the bution of the classical hydrophobic effect41 due to the larger
selectivity for different methylation states is similar. We attri- surface area of the receptor cavity, as well as a larger number
bute this difference in affinity to a systematic error in the of favorable binding orientations.
determination of receptor concentration due to incomplete
Selectivities of A2N and A2B for different methylation states
desalting in the FA experiments that influenced the reported of Lys. meso2-A2N exhibits markedly improved selectivity for
Kd. Since analysis of ITC data is not dependent on an absolute Kme3 over all other methylation states of Lys relative to rac-
concentration of host for determination of Ka, ΔH, and ΔS, A2B. meso2-A2N binds to H3 K9me3 with 14-, 130-, and 35-fold
rather the relative ratio of host concentration to Ka and the selectivity over H3 K9me2, H3 K9me, and H3 K9, respectively
absolute guest concentration,33,34 we believe the ITC data is a (Table 3, entries 1–4). In contrast, rac-A2B was found to bind
more accurate measure of the binding affinity.
the same H3 K9me3 peptide with only 2.4-, 5.4-, and 8.3-fold
Comparison of H3 K9me3 binding to A2B versus A2N. The selectivity over H3 K9me2, H3 K9me, and H3 K9, respectively
binding of all three isomers of A2N to H3 K9me3 was studied (Table 3, entries 11–14). Thus, the deeper aromatic pocket in
(see ESI, Table S2†), but after determining that meso2-A2N A2N results in a significant improvement in selectivity.
binds the tightest and most selectively to an H3 K9me3
Comparison of ΔH and ΔS for binding of the H3 Kme1–3
peptide (Table 2), further studies focused on this isomer. peptides provides some insight into the observed selectivity.
meso2-A2N was found to bind H3 K9me3 with 300 nM affinity, The driving force for meso2-A2N binding methylated Lys in the
as compared to the 2.6 μM affinity of rac-A2B (Table 3, entries 1 H3 K9 series is a favorable enthalpic term that is fairly con-
and 11). This amounts to a 1.3 kcal mol−1 difference in affinity stant for the three guests, Kme1–3. The selectivity for Kme3
arising from the introduction of an additional aromatic ring. arises primarily from a decrease in the entropic penalty of
This value is consistent with previous measurements in cyclo- binding with increasing methylation on Lys. A similar trend
phanes and β-hairpins, which showed that the cation–π inter- has been seen with a beta-hairpin system that investigated the
action with quaternary ammonium ions can contribute role of Lys methylation on cation–π interactions.38,42 There are
∼0.5–1.1 kcal mol−1 per aromatic ring to the binding of cat- several factors that may contribute to this entropic effect. The
ionic guests.35–38
peptide–receptor complex may have a larger number of favor-
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 7059–7067 | 7063

View Article Online
Paper
Organic & Biomolecular Chemistry
able binding conformations for Kme3 than for Kme2 and Kme. observed for H3 R8G-K9 (entry 8), indicating that Arg is not a
Additionally, greater methylation would be expected to result significant binder on its own. Furthermore, methylation of
in a larger contribution of the classical hydrophobic effect to Arg8 to any of the three methylated states found on histone
binding. Lastly, it may reflect different degrees of ordered tails (Rme, sRme2, and aRme2) leads to weaker binding com-
water molecules within the pocket upon binding different pared to unmethylated Arg, presumably due to weakening of
methylation states, since Kme and Kme2 can form hydrogen the unique interaction of meso2-A2N with unmethylated Arg8
bonds, unlike Kme3.43
and Lys9 (Table S1†).
Binding of meso2-A2N to H3 K9 does not follow the same
Taken together, these results suggest that the presence of
trend. meso2-A2N exhibits a tighter affinity for H3 K9 than R8 results in a different binding mechanism of H3 K9 to
H3 K9me (Table 3, entries 3 and 4). The binding of H3 K9 is meso2-A2N that is much less entropically costly than those with
considerably less exothermic than binding to the methylated methylated K9. Because we observe similar selectivities for
residues, thus its tighter affinity over H3 K9me can be attribu- Kme3 over Kme2 and Kme despite the R8G mutation, the 250-
ted to more favorable entropy of binding (compare entries 3 fold selectivity over K in the R8G series of peptides more accu-
and 4). This suggests a change in mechanism of binding, such rately represents the selectivity of meso2-A2N in the absence of
as H3 K9 binding to the exterior of the receptor via electro- other neighboring interactions.
static interactions between the carboxylates and both R8 and
The role of R8 was also investigated in the binding of rac-
K9. The favorable entropy of binding is consistent with both A2B to K9me3 and K9. In this case, mutation of R8 has about
the fact that there are multiple possible orientations for the same effect on binding to K9me3 or K9: loss of R8 results
binding and that electrostatic interactions with both in a 1.1 kcal mol−1 decrease in binding, regardless of the
ammonium and guanidinium groups have been shown to be methylation state of Lys (compare entries 11 and 15 to entries
entropically favorable in other systems.44,45 The role of R8 is 14 and 16).
explored further below.
Comparison of the enthalpy and entropy of binding of
Binding of rac-A2B to methylated Lys in the H3 K9meX (X = H3 K9me3 and H3 R8G-K9me3 with A2N or A2B provides
0–3) peptides is also driven by a release of heat that overcomes additional insights into the role of Arg in the presence of
the entropic cost of binding (Table 3, entries 11–14). However, methylated Lys. With both receptors, mutation of Arg8 to Gly
in contrast to meso2-A2N, the selectivity for H3 K9me3 is not results in a more favorable enthalpy of binding by 1.1–1.4 kcal
purely entropy driven; instead it arises from a combination of mol−1 and a less favorable entropy of binding by 2.2 kcal mol−1
enthalpic and entropic effects. With increasing methylation up (compare entries 1 to 5 and 11 to 15). Thus, the contribution
to Kme2, the binding to rac-A2B becomes more exothermic, but of Arg to binding is entropic, not enthalpic. This may suggest
more entropically disfavored, thus displaying typical enthal- additional contributions, such as Arg stacking with the aro-
phy–entropy compensation.46,47
matic rings on the exterior of the receptor, which may release
Investigation of electrostatic contributions in the H3 K9 water molecules and strengthen the electrostatic interaction
peptide. Intrigued by the peculiar tighter binding of meso2- with the carboxylates on each monomer (Fig. 5). Evidence
A2N to H3 K9 over H3 K9me, we mutated the neighboring Arg8 for this mode of binding comes from several model
to Gly to see what impact the nearby charge has on binding systems.44,45,49 Further investigation into the mechanism of
to Kmex (Table 3, entries 5–8). Upon mutation, we observed this interaction is underway.
∼4-fold weaker binding to H3 R8G-K9me3 as compared to the
Comparison of H3 K9me3 and H3 K36me3. To better under-
unmutated H3 K9me3 peptide, amounting to a loss of about stand the impact of the surrounding sequence on the recog-
0.9 kcal mol−1. As the affinities decreased in the series, we nition of Kme3 over unmodified Lys by A2N, we measured the
were unable to achieve c-values greater than the accepted binding of meso2-A2N to trimethylated and unmethylated
minimum of 1, thus the Kd values reported in these situations H3 K36 peptides, which have the same net charge but contain
are approximate and
a thermodynamic analysis is not a neighboring Lys in place of Arg (see Tables 2 and 3, entries
made.33,48 Nonetheless, the selectivity for H3 K9me3 over 19–22). The binding affinity of meso2-A2N to H3 K36me3 was
H3 K9me2 and H3 K9me is relatively unaffected by the R8G identical to that of H3 K9me3, validating that, with the exception
mutation (compare entries 1–3 to entries 5–7). In contrast,
mutation of R8 has an immense effect on binding to the
unmethylated K9, with a decrease in binding affinity of more
than 30-fold (>2 kcal mol−1). This results in a much improved
selectivity for Kme3 over K of >250-fold in this mutant series.
Comparing H3 K9 to H3 R8G-K9 (entries 4 and 8), the differ-
ence in binding affinity amounts to at least 2 kcal mol−1, com-
pared to about 1 kcal mol−1 for H3 K9me3 versus H3
R8G-K9me3 (entries 1 and 5). Thus, R8 contributes more to
binding of the unmethylated Lys than to any of the H3 Kme1–3
Fig. 5 Computational model of the interaction of a guanidinium group
peptides. It is important to note, however, that mutation of K9,
with both the carboxylates and aromatic ring of the exterior of an A or N
giving H3 R8-K9G (entry 9) results in similar weak binding monomer.
7064 | Org. Biomol. Chem., 2014, 12, 7059–7067
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 4 Comparison of binding affinities and selectivities of synthetic receptors for Kme3 peptides
Peptide
charge
Selectivity
(Kme3/K)
Entry
Host
Histone 3 peptide
Kd (μM)
Reference
1
2
3
4
5
6
rac-A2B
meso2-A2N
CX4
Ac-WGGG-QTARKme3STG-NH2
Ac-WGGG-QTARKme3STG-NH2
Ac-TARKme3STGY-NH2
H-ARTKQTARKme3STGY-NH2
H-ARTKQTARKme3STGY-NH2
H-ARTKQTARKme3STGY-NH2
+2
+2
+2
+5
+5
+5
2.6a
0.3a
7.2b
0.17c
0.19c
4.8c
8
35
This work
This work
Ref. 18
Ref. 19
Ref. 19
14
CX4
NRd
NRd
NRd
−
CX4ArCO2
CX4ArBr
Ref. 19
a 10 mM borate buffer, pH 8.5, 26 °C. b 40 mM phosphate buffer, pH 7.4, 30 °C. c 10 mM phosphate buffer, pH 7.4, 25 °C. d NR = not reported.
Conclusions
In summary, we have used iterative design coupled with DCC
to optimize a receptor for recognition of Kme3, resulting in a
300 nM binder for H3 K9me3 with 10-fold improvement in
binding affinity and a 5-fold improvement in selectivity over
Kme2. Further, meso2-A2N is the tightest and most selective
receptor for Kme3 in the context of a peptide reported to date.
Fig. 6 Structure of other reported hosts for KMe3 in the context of
peptides.
NMR data indicate that the Kme3 sidechain binds inside the
aromatic pocket, while the peptide backbone is not involved in
binding. The improved selectivity over the original receptor
arises from both more favorable enthalpy and entropy of
binding, while the improved selectivity over the lower methyl-
of basic residues, the surrounding sequence does not have a
significant impact on affinity. Interestingly, however, the
ation states of Lys arise from more favorable entropy. This
selectivity for H3 K36me3 over H3 K36 is very similar to that
work demonstrates the utility of DCC coupled with iterative
observed for the mutated H3 R8G-K9 peptide series, which has
a +1 charge (compare entries 19 and 20 to entries 5 and 8).
This supports the fact that A2N recognizes Kme3 with approxi-
mately 250-fold selectivity over K and suggests that a neighbor-
ing Arg can interact with the receptor in a unique manner.
Comparison to other synthetic receptors for Kme3. Several
synthetic receptors that bind Kme3 either as a single amino
redesign for generating new receptors with affinity and selecti-
vity necessary for biological applications and provides new
insights into the driving force for achieving both affinity and
selectivity for this class of modified amino acids in aqueous
solution. Molecular recognition in water is an ongoing chal-
lenge in supramolecular chemistry, but this work demon-
strates the strength of iterative redesign coupled with DCC for
acid23,26 or within the context of a histone tail peptide18,21
meeting this challenge.50
have been reported to date. Because the zwitterionic nature of
the amino acid influences binding in ways that are not relevant
to recognition of PTMs in proteins, only comparison to recep-
tors that bind Kme3 in the context of peptides is made here
(Table 4). It is clear that all receptors reported to date are influ-
Experimental
enced by the net charge of the peptide, such that significantly A detailed synthetic procedure for the preparation of monomer
tighter binding can be achieved with more basic peptides N can be found in the ESI.† All dynamic combinatorial
(compare Table 4, entries 3 and 4, for example). A careful ana- libraries were prepared by dissolving monomers at 2.5 mM
lysis of the effect of these nonspecific electrostatic interactions and peptide guests at 2.5–5 mM (equal to total monomer con-
on selectivity over the unmodified peptide has not been fully centration) in 50 mM sodium borate buffer (pH 8.5). DCLs
investigated for any systems. Nonetheless, comparing binding were monitored at various time points on an Agilent Rapid
to peptides of the same net charge, meso2-A2N demonstrates Resolution LC-MS system equipped with an online degasser,
the tightest binding affinity and highest selectivity over the binary pump, autosampler, heated column compartment, and
unmethylated state reported to date (Table 4, entries 1–3). diode array detector. All separations were performed using gra-
−
Interestingly, the extra aromatic ring in CX4ArCO2 relative to dients between water (A) and acetonitrile (B) containing 5 mM
CX4 does not provide any additional affinity (Table 4, entries 4 NH4OAc at pH 5 on a Zorbax Extend C18 (2.1 × 50 mm,
and 5), unlike the additional aromatic ring in A2N relative to A2B 1.8 μm) column. Using 1 μL injections, libraries were moni-
(Table 4, entries 1 and 2). The rigid nature of the rings in N as tored at various time points with the following gradient:
well as the methine linkers between the rings (versus the sulfona- 3–27% B from 0–2 min, 27–29.2% B from 2–12 min, then
−
mide linker in CX4ArCO2 and CX4ArBr) may be important in 100% B for 5 min. The MS was performed using a single
providing additional binding affinity (Fig. 2 and 6, respectively).
quad mass spectrometer. Mass spectra (ESI−) were acquired in
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 7059–7067 | 7065

View Article Online
Paper
Organic & Biomolecular Chemistry
ultrascan mode by using a drying temperature of 350 °C, a
3 S. B. Hake, A. Xiao and C. D. Allis, Br. J. Cancer, 2004, 90,
761–769.
nebulizer pressure of 45 psi, a drying gas flow of 10 L min−1
,
and a capillary voltage of 3000 V. Data analysis was performed
using the software Agilent ChemStation.
4 G. G. Wang, C. D. Allis and P. Chi, Trends Mol. Med., 2007,
13, 363–372.
A2N was prepared on a preparative scale using acetylcholine
chloride (AcCh) as a guest. The preparative DCLs were pre-
pared at 2 mM A and N and 10 mM AcCh in 50 mM sodium
borate buffer (pH 8.5) and were allowed to equilibrate for five
days before purification on an Atlantis PrepT3 5 μm 10 ×
100 mm C18 column in NH4OAc buffered solvents. Extended
lyopholization removed the majority of NH4OAc salts for an
accurate determination of the extinction coefficient.
5 H.-C. Tsai and S. B. Baylin, Cell Res., 2011, 21, 502–517.
6 M. Rodríguez-Paredes and M. Esteller, Nat. Med., 2011, 17,
330–339.
7 W. Timp and A. Feinberg, Nat. Rev. Cancer, 2013, 13,
497–510.
8 S. D. Taverna, H. Li, A. J. Ruthenburg, C. D. Allis and
D. J. Patel, Nat. Struct. Mol. Biol., 2007, 14, 1025–1039.
9 D. E. Schones and K. Zhao, Nat. Rev. Genet., 2008, 9,
179–191.
All 1D and 2D NMR experiments were performed using a
Bruker 400 MHz or Bruker 600 MHz instrument, as noted. 10 Y. Zhao and O. N. Jensen, Proteomics, 2009, 9, 4632–4641.
Data analysis was performed using Topspin 3.1 software. VT 11 S. B. Rothbart, K. Krajewski, B. D. Strahl and S. M. Fuchs,
1D NMRs were collected on a Bruker 500 MHz instrument
using cooled or heated nitrogen gas to control the temperature 12 Z. Su, M. D. Boersma, J.-H. Lee, S. S. Oliver, S. Liu,
of the sample. NMR binding experiments were performed on a B. A. Garcia and J. M. Denu, Epigenetics Chromatin, 2014, 7, 7.
Methods Enzymol., 2012, 107–135.
Bruker 600 MHz instrument at 25 °C in 10 mM borate 13 S. Thorpe, Nature, 2007, 447, 741–744.
buffererd D2O (pH 8.67). Concentration determination of 14 I. Bock, A. Dhayalan, S. Kudithipudi, O. Brandt, P. Rathert
Ac-Kme3G-NH2 was performed using DSS as an internal standard.
and A. Jeltsch, Epigenetics, 2011, 6, 256–263.
A 600 μM peptide stock solution was used to dissolve lyophi- 15 T. A. Egelhofer, A. Minoda, S. Klugman, K. Lee,
lized meso2-A2N or rac-A2B. 1D spectra were collected with 128
scans. Proton assignments were made using TOCSY analysis.
All ITC titrations were performed using a MicroCal Auto-
iTC200 at 26 °C. Data analysis was performed using the built
in Origin 7 software using a one site binding model. Unless
otherwise noted, titrations were performed in triplicate. A
10 mM pH 8.5 sodium borate buffer was used for all experi-
ments. All concentrations were determined using a Nano-
P. Kolasinska-Zwierz, A. A. Alekseyenko, M.-S. Cheung,
D. S. Day, S. Gadel, A. A. Gorchakov, T. Gu,
P. V. Kharchenko, S. Kuan, I. Latorre, D. Linder-Basso,
Y. Luu, Q. Ngo, M. Perry, A. Rechtsteiner, N. C. Riddle,
Y. B. Schwartz, G. A. Shanower, A. Vielle, J. Ahringer,
S. C. R. Elgin, M. I. Kuroda, V. Pirrotta, B. Ren, S. Strome,
P. J. Park, G. H. Karpen, R. D. Hawkins and J. D. Lieb, Nat.
Struct. Mol. Biol., 2011, 18, 91–93.
Drop2000 with a xenon flash lamp, 2048 element linear silicon 16 S. M. Fuchs, K. Krajewski, R. W. Baker, V. L. Miller and
CCD array detector, and 1 mm path length. Peptides were D. Brian, Curr. Biol., 2012, 21, 53–58.
desalted using Thermo Scientific polyacrylamide 1800 MWCO 17 S. Nishikori, T. Hattori, S. M. Fuchs, N. Yasui, J. Wojcik,
desalting columns. ∼0.5–3 mM solutions of peptide were
titrated into ∼20–200 μM solutions of A2N (all isomers were
A. Koide, B. D. Strahl and S. Koide, J. Mol. Biol., 2012, 424,
391–399.
studied, see ESI† for data on the rac- and meso1-isomers) or 18 K. D. Daze, T. Pinter, C. S. Beshara, A. Ibraheem,
rac-A2B using 2 μL injections every 3 minutes. Heats of dilution
of peptides were subtracted prior to analysis in Origin.
S. A. Minaker, M. C. F. Ma, R. J. M. Courtemanche,
R. E. Campbell and F. Hof, Chem. Sci., 2012, 3, 2695–2699.
19 M. Florea, S. Kudithipudi, A. Rei, M. J. Gonzalez-Alvarez,
A. Jeltsch and W. M. Nau, Chem. – Eur. J., 2012, 3521–3528.
20 S. A. Minaker, K. D. Daze, M. C. F. Ma and F. Hof, J. Am.
Chem. Soc., 2012, 11674–11680.
Acknowledgements
21 H. F. Allen, K. D. Daze, T. Shimbo, A. Lai, C. A. Musselman,
J. K. Sims, P. A. Wade, F. Hof and T. G. Kutateladze,
Biochem. J., 2014, 459, 505–512.
22 L. A. Ingerman, M. E. Cuellar and M. L. Waters, Chem.
Commun., 2010, 46, 1839–1841.
We gratefully acknowledge funding from the W. M. Keck foun-
dation for this work. This material is based in part upon work
supported by the National Science Foundation under grant no.
CHE-1306977 and also the National Science Foundation
Graduate Research Fellowship to N.K.P. under grant no.
DGE-1144081.
23 K. D. Daze, M. C. F. Ma, F. Pineux and F. Hof, Org. Lett.,
2012, 14, 1512–1515.
24 A. L. Whiting and F. Hof, Org. Biomol. Chem., 2012, 10,
6885–6892.
25 K. D. Daze and F. Hof, Acc. Chem. Res., 2013, 46, 937–945.
26 M. A. Gamal-Eldin and D. H. Macartney, Org. Biomol.
Chem., 2013, 11, 488–495.
27 L. I. James, J. E. Beaver, N. W. Rice and M. L. Waters, J. Am.
Chem. Soc., 2013, 135, 6450–6455.
Notes and references
1 C. Martin and Y. Zhang, Nat. Rev. Mol. Cell Biol., 2005, 6,
838–849.
2 T. Kouzarides, Cell, 2007, 128, 693–705.
7066 | Org. Biomol. Chem., 2014, 12, 7059–7067
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
28 S. A. Jacobs and S. Khorasanizadeh, Science, 2002, 295, 40 M. V. Rekharsky, T. Mori, C. Yang, Y. H. Ko, N. Selvapalam,
2080–2083.
H. Kim, D. Sobransingh, A. E. Kaifer, S. Liu, L. Isaacs,
W. Chen, S. Moghaddam, M. K. Gilson, K. Kim and
Y. Inoue, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 20737–
20742.
29 P. T. Corbett, J. Leclaire, L. Vial, K. R. West, J.-L. Wietor,
J. K. M. Sanders and S. Otto, Chem. Rev., 2006, 106, 3652–3711.
30 P. T. Corbett, J. K. M. Sanders and S. Otto, Chem. – Eur. J.,
2008, 14, 2153–2166.
41 C. Tanford, Science, 1978, 200, 1012–1018.
31 D. H. Hua, M. Tamura, X. Huang, H. A. Stephany, 42 R. M. Hughes, M. L. Benshoff and M. L. Waters, Chem. –
B. A. Helfrich, E. M. Perchellet, B. J. Sperfslage, J.-P. Perchellet, Eur. J., 2007, 13, 5753–5764.
S. Jiang, D. E. Kyle and P. K. Chiang, J. Org. Chem., 2002, 43 F. Biedermann, V. D. Uzunova, O. A. Scherman, W. M. Nau
67, 2907–2912.
and A. De Simone, J. Am. Chem. Soc., 2012, 134, 15318–
32 S. Otto, R. L. E. Furlan and J. K. M. Sanders, Science, 2002,
297, 590–593.
33 T. Wiseman, S. Williston, J. F. Brandts and L. N. Lin, Anal.
Biochem., 1989, 179, 131–137.
15323.
44 S. E. Thompson and D. B. Smithrud, J. Am. Chem. Soc.,
2002, 124, 442–449.
45 S. L. Tobey and E. V. Anslyn, J. Am. Chem. Soc., 2003, 125,
14807–14815.
34 J. Tellinghuisen, J. Phys. Chem. B, 2005, 109, 20027–20035.
35 H. Schneider, T. Matter and S. Simova, J. Chem. Soc., Chem. 46 E. Grunwald and C. Steel, J. Am. Chem. Soc., 1995, 5687–
Commun., 1989, 580–581. 5692.
36 P. C. Kearney, L. S. Mizoue, R. A. Kumpf, J. E. Forman, 47 M. V. Rekharsky and Y. Inoue, Chem. Rev., 1998, 98, 1875–
A. Mccurdy and D. A. Dougherty, J. Am. Chem. Soc., 1993,
115, 9907–9919.
37 C. D. Tatko and M. L. Waters, Protein Sci., 2003, 2443–2452.
1918.
48 W. Turnbull and A. Daranas, J. Am. Chem. Soc., 2003,
14859–14866.
38 R. M. Hughes and M. L. Waters, J. Am. Chem. Soc., 2005, 49 X. Wang, J. Post, D. K. Hore and F. Hof, J. Org. Chem., 2014,
127, 6518–6519. 79, 34–40.
39 S. L. Hauser, E. W. Johanson, H. P. Green and P. J. Smith, 50 G. V. Oshovsky, D. N. Reinhoudt and W. Verboom, Angew.
Org. Lett., 2000, 2, 3575–3578.
Chem., Int. Ed., 2007, 46, 2366–2693.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 7059–7067 | 7067