Structure optimization of lipopeptide assemblies for aldol reactions in an aqueous medium

Article abstract of DOI:10.1039/d1cp01060c

Four amphiphilic peptides were synthesized, characterized, and evaluated regarding their efficiency in the catalysis of direct aldol reactions in water. The lipopeptides differ by having a double lipid chain and a guanidinium pyrrole group functionalizing one Lys side chain. All the samples are composed of the amino acids l-proline (P), l-arginine (R), or l-lysine (K) functionalized with the cationic guanidiniocarbonyl pyrrole unit (GCP), l-tryptophan (W), and l-glycine (G), covalently linked to one or two long aliphatic chains, leading to surfactant-like designs with controlled proline protonation state and different stereoselectivity. Critical aggregation concentrations (cac) were higher in the presence of the GCP group, suggesting that self-assembly depends on charge distribution along the peptide backbone. Cryogenic Transmission Electron Microscopy (Cryo-TEM) and Small Angle X-ray Scattering (SAXS) showed a rich polymorphism including spherical, cylindrical, and bilayer structures. Molecular dynamics simulations performed to assess the lipopeptide polymorphs revealed an excellent agreement with core-shell arrangements derived from SAXS data and provided an atomistic view of the changes incurred by modifying head groups and lipid chains. The resulting nanostructures behaved as excellent catalysts for aldol condensation reactions, in which superior conversions (>99%), high diastereoselectivities (ds = 94?:?6), and enantioselectivities (ee = 92%) were obtained. Our findings contribute to elucidate the effect of nanoscale organization of lipopeptide assemblies in the catalysis of aldol reactions in an aqueous environment.

Full text of DOI:10.1039/d1cp01060c

PCCP
View Article Online
View Journal | View Issue
PAPER
Structure optimization of lipopeptide assemblies
for aldol reactions in an aqueous medium†
Cite this: Phys. Chem. Chem. Phys.,
2021, 23, 10953
Bruna M. Soares,a Pedro T. Sodre,a Andrea M. Aguilar,b Barbara B. Gerbelli,
a
´
c
Juliane N. B. D. Pelin,a Karina B. Arguello,a Emerson R. Silva,
¨
Marcelo A. de Farias,d Rodrigo V. Portugal,d Carsten Schmuck,
‡
e
a
a
Maurıcio D. Coutinho-Neto
and Wendel A. Alves
*
´
Four amphiphilic peptides were synthesized, characterized, and evaluated regarding their efficiency in
the catalysis of direct aldol reactions in water. The lipopeptides differ by having a double lipid chain and
a guanidinium pyrrole group functionalizing one Lys side chain. All the samples are composed of the
amino acids L-proline (P), L-arginine (R), or L-lysine (K) functionalized with the cationic
guanidiniocarbonyl pyrrole unit (GCP), L-tryptophan (W), and L-glycine (G), covalently linked to one or
two long aliphatic chains, leading to surfactant-like designs with controlled proline protonation state and
different stereoselectivity. Critical aggregation concentrations (cac) were higher in the presence of the
GCP group, suggesting that self-assembly depends on charge distribution along the peptide backbone.
Cryogenic Transmission Electron Microscopy (Cryo-TEM) and Small Angle X-ray Scattering (SAXS)
showed a rich polymorphism including spherical, cylindrical, and bilayer structures. Molecular dynamics
simulations performed to assess the lipopeptide polymorphs revealed an excellent agreement with
core–shell arrangements derived from SAXS data and provided an atomistic view of the changes
incurred by modifying head groups and lipid chains. The resulting nanostructures behaved as excellent
catalysts for aldol condensation reactions, in which superior conversions (499%), high diastereoselectivities
(ds = 94 : 6), and enantioselectivities (ee = 92%) were obtained. Our findings contribute to elucidate the
effect of nanoscale organization of lipopeptide assemblies in the catalysis of aldol reactions in an aqueous
environment.
Received 9th March 2021,
Accepted 20th April 2021
DOI: 10.1039/d1cp01060c
rsc.li/pccp
study of self-organization phenomena occurring in either living
organisms or catalytic process.1–7
Introduction
Investigations on self-assembling properties of amphiphilic
Aldol reactions are among the most important and investigated
peptides have emerged in recent years as an active and diversified transformations in organic synthesis because a new C–C bond and
field involving either fundamental or application-oriented a b-hydroxy carbonyl structure are formed from two carbonyl
research in biotechnology and materials science. These materials compounds.8 Asymmetric aldol reactions are characterized
exhibit intrinsic multi-level organization, and this structural by the formation of new stereogenic carbons, increasing the
multiplicity generates several types of intra- and intermolecular reactions’ selectivity levels.9
interactions, making them extremely attractive as a model for the
The efficiency of peptides containing L-proline as an organo-
catalyst in aldol condensation reactions has been known for over
the past half-century.10–12 This methodology was revisited in
2000 when this amino acid was proved to be very efficient at
promoting organocatalyzed aldol reactions.13,14 These studies
are important milestones for the development of new organo-
catalysts containing proline and the corresponding application
in several organic reactions.15,16
a
ˆ
Centro de Ciencias Naturais e Humanas, Universidade Federal do ABC,
´
09210-580, Santo Andre, Brazil. E-mail: wendel.alves@ufabc.edu.br
b
c
ˆ
´
ˆ
Instituto de Ciencias Ambientais, Quımicas e Farmaceuticas,
˜
Universidade Federal de Sao Paulo, 09972-270, Diadema, Brazil
´
˜
Departamento de Biofısica, Universidade Federal de Sao Paulo, 04023-062,
˜
Sao Paulo, Brazil
d Brazilian Nanotechnology National Laboratory, CNPEM, 13083-970, Campinas,
Brazil
Previous studies have shown that the supramolecular
hydrophobic-amphiphilic approach is responsible for the high effi-
ciency of L-proline-derived peptides as organocatalysts in asymmetric
aldol reactions.17–22 Proline-based compounds were also shown to be
catalysts in water with the aid of polymer or surfactant.23,24
e Institute for Organic Chemistry, University of Duisburg-Essen, 45117, Essen,
Germany
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
d1cp01060c
‡ This article is dedicated to the memory of Professor Carsten Schmuck.
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10953

View Article Online
Paper
PCCP
Recently, we studied the effect of substitution of the ester by reduces charge repulsion while enabling strong interactions
an amide linkage between the hydrophobic lipid chain and the between side chains, thus assisting stabilization of
PRW tripeptide headgroup (where, P: L-proline, R: L-arginine, aggregates.27 We synthesized two series of lipopeptides, in both
W: L-tryptophan).25 Both lipopeptides have been shown to presence and absence of GCP groups, and studied their self-
self-assemble into micelles above a critical aggregation concen- assembly and their effects on directly catalyzing asymmetric
tration, detected through intrinsic tryptophan fluorescence or aldol reactions of cyclohexanone and p-nitrobenzaldehyde
using an added fluorophore probe (pyrene, the fluorescence of in water.
which is sensitive to the hydrophobic environment). Aldol
Molecular dynamics simulations was used previously19 to
reactions using cyclohexanone and p-nitrobenzaldehyde assist characterization of PRW-O-C16 lipopeptide structure and
showed that PRW-NH-C16 has an excellent anti/syn diastereomeric understand its role on the pre-concentration of reactants. Here,
ratio (93 : 7) and a high enantiomeric excess (ee) of 88%. These we have extended these analyses to understand the differences
values are better than those observed for PRW-O-C16 under the between the distinct micelles produced from self-assembling
same conditions (anti/syn = 91 :9 and ee = 71%).19 Here, the linker PRW-O-C16, PRW-NH-C16, PRWG-C18 (1), and PRWG-(C18)2 (2).
group had an influence that is potentially due to differences in the We have also investigated the effect of adding a Lys-GCP group
local conformation around the catalytic site and/or the altered over the structural properties of the micelle (compounds 3 and
polarization of the amide vs. ester linkage. Also, for both systems, 4). Moreover, quantum mechanical calculations were carried to
low correlations were observed in the absence of water. This effect investigate a possible role of the micro-environment provided
can be associated with the lower organization of the lipopeptide by the micellar arrangement at controlling proline protonation.
molecules due to weakening of hydrophobic interactions
that drive self-assembly into micelles. We have also found that
the L-proline residue influences the lipopeptide self-assembly,
Experimental
considering RWG and PRWG residues attached to one or two
aliphatic octadecyls chains.25 The organocatalyzed aldol efficiency
was increased by water availability, achieving almost 95%
conversion and excellent diastereoselectivity (93 : 7). Also, it was
observed an enhancement of conversion ratio upon the growth of
proline-lipopeptide content, which formed micelles with a core
radius of 3 nm and shell thickness of 2 nm.
In the present work, we carried out an experimental and
theoretical study on the influence of replacing arginine residues
by a synthetic arginine mimetic, the guanidiniumcarbonyl
pyrrole unit (GCP), in PRWG-C18 and PRWG-(C18)2 lipopeptides
(Fig. 1). GCP is an efficient binding site for oxoanions in polar
solutions.26 Owing to its weak basicity (pKa = 7), the only partially
protonated state within the peptide composition efficiently
Lipopeptide synthesis
All the experimental procedures for synthesizing the lipopeptides
(1–4) are described in detail in the ESI.† Also, the mass spectro-
scopy spectra can be seen in Fig. S1 (ESI†).
General information
All solutions were prepared with water purified by the Direct-Q
System, Millipore, with a resistivity of 18.2 mO cmÀ1 (at 25 1C)
and TOC below 10 ppb. The reagents and solvents used were all
of the analytical purity.
Fluorescence spectroscopy
Lipopeptide solid was firstly dissolved in 20 ml of HFIP
(1,1,1,3,3,3-hexafluoro-2-propanol) and then water solutions
(1.0 Â 10À7 to 0.1 wt%) were prepared. After one day of rest,
the fluorescence of the systems was evaluated irradiated with
lexc = 280 nm, and the emission spectrum were investigated in
the range 300 nm r lem r 460 nm. Emission spectra were
obtained using fluorescence spectrophotometer, Cary Eclypse
(Varian), model and quartz cuvette, SUPRASIL 10 Â 10 mm,
(Hellma Analytics).
Cryogenic transmission electron microscopy (Cryo-TEM)
Cryo-TEM data were conducted at the LNNano, Brazilian Nano-
technology National Laboratory (CNPEM, Campinas, Brazil).
Cryo-TEM images of 1 wt% of the lipopeptides were performed.
The grid vitrification was made using Vitrobot Mark IV (Thermo
Fisher Scientific, USA) instrument, using blot time: 1.0 or 2.5 s;
blot force: À5 s, blot wait: 20 s, with temperature and humidity
of 22 1C and 100%, respectively. For this, 3.0 mL of the sample
were deposited on the lacey carbon-coated copper grid (Ted
Pella, USA) with 300 mesh, which were submitted to glow
discharge procedure using EasiGlow (Pelco, USA), using nega-
tive charge. The grids were firstly immersed in liquid ethane,
Fig. 1 Molecular structures of the lipopeptides PRWG-C18H37 (1), PRWG-
(C18H37)2 (2), PK(GCP)WG-C18H37 (3) and PK(GCP)WG-(C18H37)2 (4).
10954 | Phys. Chem. Chem. Phys., 2021, 23, 10953–10963
This journal is © the Owner Societies 2021

View Article Online
PCCP
Paper
and then they were kept in liquid nitrogen until the moment of used the standard RESP35 charge scheme implemented in
analysis. The images were obtained by transmission electron Ambertools32,36 to derive charges for the alkyl section using
microscope model TALOS F200C (Thermo Fisher Scientific, an acetyl cap. We employed the Hartree–Fock method and a
USA), operating at 200 kV, with CMOS Ceta 16M camera. The 6-31G* basis set in all RESP charge derivations. Tip3P37 water
pixel size of the images varied from 0.161 nm to 0.413 nm. and ClÀ counter ions were added to the system to ensure
ImageJ program was used to calculate the size of particles.
neutrality. For simplicity’s sake ClÀ was used instead of tri-
fluoroacetic acid (TFA) as the counter ion in our simulations.
Packmol38,39 was used to provide initial configurations for
Small angle X-ray scattering (SAXS)
SAXS data were collected using the SAXS 1 beamline at the all micellar systems. Initial conditions for systems (1), (2), and
LNLS, Brazilian Synchrotron Light Laboratory (CNPEM, (4) were selected to packing into a spherical object with a final
Campinas, Brazil). The integrated intensity is displayed as a diameter of circa 7 to 8 nm after minimization. The alkyl chains
function of the momentum transfer modulus q, where q is were packed into an inner sphere, with the peptide part
defined as q = 4p sin(y)/l, which 2y is the scattering angle and l pointing outwards. System (3) was packed into a cylinder with
is the radiation wavelength. The measured q range was a diameter of circa 6 nm. Ions and water were added using leap
0.0150 ÅÀ1 r q r 0.45 ÅÀ1 with radiation energy of 8 keV from Ambertools with a water shell of at least 1.2 nm. We have
(l = 1.5498 Å), using sample holders for liquid samples used it in all simulated systems 160 lipopeptide units. The
available on the beamline at room temperature. Ten frames geometric parameters that control packing for PRWG-C18 and
of 60 seconds each were recorded to avoid radiation damage. PRWG-(C18)2 are close enough to PRW-O-C16. Hence, we opted
Afterward, data treatment was performed by subtracting the to perform our simulations using the same packing number as
solvent normalized intensity from normalized intensity of in our previous work19 to simplify comparisons between the
particles in solution, using Fit2D software.28 Data fitting was different systems.
performed using SASFit program.29
Six simulation systems were considered, namely systems
PRWG-C18 (1), PRWG-(C18)2 (2), PK(GCP)WGC18H37 (3), PK(GCP)WG-
(C18H37)2 (4), PRW-N-C16 and PRW-O-C16. PRW-O-C16 has been
Aldol reactions
The catalytic efficiency of the lipopeptides was monitored by the previously investigated experimentally and theoretically.19
direct aldol reaction using cyclohexanone and p-nitrobenzaldehyde. However, due to small changes in the employed protocol and
The reactions were performed using different catalyst amounts simulation time, we have replicated our original simulation for
of 1.0, 2.5, and 5.0 mol%. Considering the 5 mol% catalyst comparison purposes.
solutions, 100 ml of cyclohexanone (0.96 mmol, 12 equivalents),
Details of the simulation protocol used in our simulations
3.0 mg of catalyst, 12.0 mg of p-nitrobenzaldehyde (80 mmol, are described in the ESI.† A production of 100 ns is performed
1 equivalent), and 200 ml of water (2Â cyclohexanone volume) for each of the simulated systems using the GPU implementation
were used. Firstly, a small amount of HFIP was added to the of AMBER 18.40,41 We used a Berendsen42 barostat with a 1.0 ps
solid lipopeptide to break the secondary structures, and after relaxation time and pressure equals 1 atm, a Langevin43 thermo-
the water was included. The systems were kept on the cupboard stat with a 2 psÀ1 collision frequency, and a timestep of 0.02 fs.
for one day to evaporate all the HFIP and promote the lipo- Shake was used to constrain the hydrogen to their equilibrium
peptide self-assembly. Then the aldol reagents were added, and positions.44
the solutions were stirred at room temperature for two days.
All structural properties were analyzed using the CPPTRAJ45
The mixtures were extracted with ethyl acetate four times via program from AMBERTOOLS. Visualizations were performed
centrifugation at 4000 rpm for 5 min. The organic phase was using VMD.46 We have computed radial distribution functions
then removed on a desiccator under a high vacuum, and the (RDFs or g(r)), the radius of gyration (ROG) for all systems
samples were solubilized using deuterated chloroform. aiming to understand the differences between their supra-
NMR measurements using a (1H) Bruker Ultrashield Plus 300 molecular arrangements.
instrument were performed at 300 MHz. The yield and diaster-
eomer anti : syn ratio were calculated using the NMR spectra
Quantum mechanical calculations
obtained, for which tetramethylsilane (TMS) was used as a Quantum mechanical (QM) calculations were performed in three
reference. The enantiomeric excess (ee) values were determined distinct environments (water, cyclohexanone, and N-heptane) to
by HPLC on a Chiralpak AD-H column.
quantify the micellar environment influence on the pKas of the
L-proline amino-terminal and trifluoroacetic acid. To simplify
our calculations, we used L-proline methyl ester instead of the
Molecular dynamics simulation
We have modelled all lipopeptides using the AMBER family of whole peptide chain. DG and pKa were computed using the
force fields. The alkyl tails were modelled using Lipid1430,31 procedure suggested by Rossini.47,48 We have used the ma-def2-
and Lipid1732 force fields, while the peptide part was modelled TZVP basis set,49,50 and B3LYP functional51–53 with empirical
using the ff14SB force field.33 The peptide–lipid boundary was dispersion corrections54,55 as implemented in Orca code version
modelled using GAFF34 parameters for amides. The PRW-O-C16 4.2.156 in all calculations. We treat solvation effects using the
ester bond connecting the peptide and alkyl chain employed implicit solvent SMD (Solvation Model Density) method from
the same parameters used in Lipid11 and Lipid14. We have Truhlar and collaborators.57 Computed results used bare and
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10955

View Article Online
Paper
PCCP
micro-solvated molecules in a procedure analog to the one show a peak in the distribution at about 5.5 angstroms, within
suggested by Pliego and Riveros.58 More specifically, in the typical distances for FRET. g(R) integral is shown in red and
micro-solvation calculations, two explicit water molecules gives the average number of GCP residues in a coordination
were added in hydrogen bond configurations to each solute sphere with radius R (distance between centers of mass)
molecules. We assumed that the solvated proton is in pure water surrounding Trp. The integral value at 7 angstroms (the first
in all calculations, reflecting the equilibrium state in a micro peak) is 6.5, meaning that there are 6.5 GCP residues with
heterogeneous environment in analogy to a procedure employed distances of less than 7 angstroms on average from each Trp
by Silva.59
residue.
Critical concentrations were obtained from the intercept
curves were determined to be (3.4 Æ 0.5) Â 10À4 wt% for (1)
and (2), (1.2 Æ 0.5) Â 10À3 wt% for (3) and (2.5 Æ 0.5) Â 10À3 wt%
for (4). This enhancement suggests higher hydrophilicity of
sequences (3) and (4) than other compounds, which could be
attributed to the GCP group charge, promoting a higher inter-
action with the water medium.
Results and discussion
The self-assembly of lipopeptides was initially evaluated by
steady-state fluorescence spectroscopy, by investigating the
intensity of the intrinsic tryptophan fluorescence to obtain
critical aggregation concentrations (cac). Emission curves from
the present study are shown in Fig. S2 (ESI†), whereas the
behavior of fluorescence intensity as a function of lipopeptide
concentration is shown in Fig. 2.
The emission of indole may be blue-shifted if the group is
buried within a hydrophobic cavity, and its emission may shift
to longer wavelengths (red-shift) when the hydrogen bonding
and the subsequent exposure of the tryptophan residue to the
aqueous phase increases.60 Both peptides (3) and (4) showed
fluorescence emission at longer wavelengths (4350 nm), likely
revealing GCP residues’ influence in the self-assembly, favoring
the high water content and closer packing of the lipopeptides,
as shown by molecular dynamic simulation. Also, compound
(3) showed an emission band at B440 nm, characteristic of
fluorescence resonance energy transfer (FRET), since the GCP/
In Fig. 3, SAXS curves from aqueous solutions prepared at
different concentrations show scattering profiles characterized
by oscillations along the q-range, indicating the presence of
monodisperse nanoparticles with regular morphology. For
some compositions and concentrations, it is possible to
observe an abrupt decay at the low angle region, indicating
the presence of interference peaks associated with interparticle
correlations.62 The datasets were firstly fitted by using the
indirect Fourier transformation (IFT) method, which allows
for determining the pair distance distribution function p(r).
P(r) curves shown in Fig. S5 (ESI†) are characterized by bimodal
profiles, which indicate core–shell assemblies in solution.63
Considering the IFT approach used above, we have fitted our
data according to spherical and cylindrical shells form factors
to obtain quantitative measurements on characteristic sizes of
nanoparticles in solution (Fig. 3a and b). Interparticle correla-
tions have been described by using a hard-sphere structure
factor in the model. In the case of samples prepared with (4), a
bilayer form factor64 has been added to account for the
presence of planar assemblies in the solution (Fig. S4e, ESI†).
The values of the best fitting parameters are summarized in
Table 1. Data from solutions prepared with (1) could be
¨
indole separation distances are shorter than the typical Forster
distance (9–16 Å).61 In this case, GCP absorption at around
300 nm (see Fig. S3, ESI†) overlaps with tryptophan emission,
producing FRET at 440 nm.
For compound (3), we have computed the radial distribution
function for the tryptophan ring/GCP pair from our molecular
dynamics simulation trajectories (see Fig. S4, ESI†). Results
Fig. 2 Fluorescence data used to estimate the CAC of lipopeptides (1–4).
The intersection between the straight lines fitted to the data from different
fluorescence regimes provided the CAC estimate.
Fig. 3 SAXS data from lipopeptide solutions (a) 0.12 wt% and (b) 0.5 wt%.
The solid lines represent the fitting to the experimental data using a
spherical shell (1) and (2), cylindrical shell (3), and bilayer form factors (4).
10956 | Phys. Chem. Chem. Phys., 2021, 23, 10953–10963
This journal is © the Owner Societies 2021

View Article Online
PCCP
Paper
Table 1 Summary of structural parameters obtained from least-square fitting of SAXS data. Rin: inner radius; Rout: outer radius; t: thickness of bilayers,
Sin: standard deviation of Gaussian function used to describe the apolar region and Sout: polar region. RHS: hard-sphere radius (effective distance
between particles) and Fp: volume fraction occupied by hard spheres
Spherical shell form factor Hard-sphere structure factor Cylinder shell form factor Bilayer form factor
Compounds Conc. (mg mlÀ1
)
RIn (nm)
Rout (nm)
RHS (nm)
FP
RIn (nm)
Rout (nm)
t (nm) Sout Sin
(1)
(2)
(3)
(4)
1.25
5
10
1.25
5
10
1.25
5
10
1.25
5
1.4
2.2
—
1.8
1.7
1.8
—
—
—
1.6
1.6
1.2
2.2
2.3
—
2.8
2.5
3.7
—
—
—
2.5
3.3
3.5
15
6
—
14
10
7
—
—
—
13
10
7
0.14
0.13
—
0.12
0.19
0.14
—
—
—
0.12
0.19
0.19
—
—
—
—
—
—
1.5
1.8
—
—
—
—
—
—
—
—
—
—
2.7
3.5
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
7.2
—
—
—
—
—
—
—
—
—
—
—
2.2
—
—
—
—
—
—
—
—
—
—
—
2.0
10
satisfactorily fitted using a spherical core–shell form factor observes that the low-q range exhibits a smooth decay scaling
convoluted with a hard-sphere structure factor. In the most with qÀ1, indicating the presence of cylindrical structures
diluted sample (0.12 wt%), we found the presence of aggregates (Fig. 3a). To fit these data, we used the cylinder core–shell form
with a core radius (Rin) of 1.4 nm and a shell (Rout) radius of factor. The structural parameters obtained for this concentration
2.2 nm (Fig. 3a). It is interesting to note that the entire data were Rin = 1.5 nm and Rout = 2.7 nm. For the intermediate
range could be adjusted with a single type of nanostructure, concentration sample (0.50 wt%) again, a cylindrical model was
reinforcing the consistency of the model. For the intermediate used, increasing the size of Rin and Rout (Fig. 3b).
concentration (0.50 wt%), the form factor also consists of the
Data from lipopeptide (4) are shown in Fig. 3a. The most
spherical core–shell type, with the same Rin of the lowest diluted sample curve (0.12 wt%) profile is accurately described
concentration but with the Rout equal to 2.3 nm, indicating by spherical shell form factors revealing low polydispersity and
that growth of nanoparticles is observed in the medium. In this an internal radius of 1.6 nm. Samples at an intermediate
case, the curve profile with absence of intensity minimums concentration (0.50 wt%) are also described by the same factor
shows a higher polydispersity, which is expected since more presenting the same value for the inner radius, but the external
peptides are present in the medium and, consequently, more radius increases by 0.8 nm (Fig. 3b). For the more concentrated
particles are in equilibrium (Fig. 3b).
conditions (1.00 wt%), data fitting was carried out by adding
Moreover, it is observed that only the particle shell has two components corresponding to a bilayer and spherical form
increased with increasing concentrations. It can be an indication factor. This form factor has been used successfully to fit the
of a new conformation from a more compact to a more form factor of lipopeptide nanotape and nanosheet structures
elongated state. This could be attributed to the fact that we (Fig. S5e, ESI†).65,66 Here, the bilayer form factor revealed that
increase the interaction with the arginine lateral group promoting our samples contain planar structures with thickness equal to
such behavior. It was impossible for the more concentrated 7.2 nm, and spherical micelles with an internal radius of
sample (1.00 wt%) to make a good fitting using a simple spherical 1.2 nm and an external radius of 2.6 nm.
model (Fig. S5e, ESI†).
A noteworthy point of samples prepared with lipopeptide (4)
A decrease of the hard-sphere repulsion radius (RHS) verified is the presence of interference peaks at dilute preparations,
upon increasing concentration suggests a higher interaction indicating that interparticle correlations are present even at low
between the particles in solution.
concentrations. These findings suggest that nanoparticles, in
The curves related to lipopeptide (2) for the most diluted this case, carry higher charge densities at their outer surfaces,
sample curve (0.12 wt%) also indicated spherical shell aggregates enhancing repulsion between self-assemblies in solution. Such
(Fig. 3a), with Rin of 1.8 nm and an external radius of 2.8 nm. an effect is likely a consequence of the GCP moiety present at
The profile of the curve is consistent with the presence of regular the headgroup of lipopeptide (4).
particles with low polydispersity. For the intermediate concen-
A recent study from our group investigated the influence of
tration sample (0.50 wt%), the same size found at low concen- proline residues in mixtures containing the lipopeptides (1) and
tration was observed (Fig. 3b). For the more concentrated system (2).25 In that case, a decrease of nanoparticle thickness upon an
(1.0 wt%), the curve shape indicates greater polydispersity with increase of proline-lipopeptide content corroborated with the
the presence of larger micelles. In this case, the internal radius is experimental data obtained in this work. As proline amino acid
1.8 nm, and the external radius is 2.8 nm (Fig. S5e, ESI†).
SAXS data are significantly different for peptides (3) and (4), increase the hydrophobicity and form more compact systems.
likely revealing the influence of GCP residues in the self- Furthermore, these results showed that the GCP charge can
assembly. For the most diluted sample prepared with (3), one change the self-assembly process of the lipopeptides in water,
has a rigid side chain, its presence in lipopeptide constructs can
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10957

View Article Online
Paper
PCCP
The computed radial distribution functions (RDFs) show a
similar layered structural organization of the residues with
respect to the micellar center for all systems. As reported
previously for PRW-O-C16, arginine and catalytic proline RDFs
peak at about the same distance from the micelle center and
comprise the hydrophilic pair. At the same time, tryptophan
and glycine are closer to the hydrophobic core. For (1), C1–C18
g(r), where C18 is the last carbon atom of the alkyl chain, peaks
at 2.33 nm while the g(r) for the C1-Pro pair peaks at 3.15 nm.
For PRWG-(C18)2, the peak values for C1–C18 and C1-Pro g(r)s
are 2.32 nm and 3.17 nm, illustrating the similarities between
these two systems. RDF between the C1 alkyl carbon and water
oxygen atoms revealed a drop in water content at about 4.5 nm
from the hydrophobic core. At the peak of the proline g(r) for (1)
(at r = 3.15 nm), the C1-Wat(O) distribution has about 61% of its
bulk value (g(r) = 0.61), pointing to an intermediate environment in
terms of water content where catalysis takes place. Interestingly,
for the PRWG-(C18)2 micelle (2), the value for the C1-Wat(O)
distribution at the peak of the proline g(r) has only 52% of its
bulk value, indicating a stronger hydrophobic nature for this
system.
Fig. 4 Cryo-TEM images of 1 wt% water solutions of (1–4). Lipopeptide
self-assembly in (1) and (2) micelles spherical, (3) micelles cylindrical, and
(4) planar structures.
The radius of gyration values (ROG) shown in Table S2 (ESI†)
for proline can be compared to SAXS results from Table 1. The
total radio (Rin + Rout) determined experimentally agree with the
ROG for the last residue, proline, being 3.78 nm and 4.22 nm
in which polymorphism can influence hydrophilicity and com- for (1) and (2), respectively. This can be compared to 3.6 nm
pression effects. To shed light on the atomistic interactions and 4.5 nm from SAXS results from Table 1. This result
that characterize the different structures, we have performed supports the choice for the aggregation number used to con-
molecular dynamics simulations for (1) and (2). After that, struct the micellar model systems, despite having similar RDFs.
we compared it to previously investigated systems, namely The larger ROG values for (2) result from its bulkier
19
PWR-O-C16 and PWR-NH-C16
.
hydrophobic core.
Cryo-TEM experiments were performed to assess the
RDFs for PRW-O-C16 and PRW-NH-C16 systems, shown in
morphology in the real space of lipopeptides (1–4) above the Fig. S10 (ESI†), present a similar pattern to (1) and (2). ROG for
cac. We determine the size of the particles using Image-J the last residue, proline, for PRW-O-C16 and PRW-NH-C16 being
program. Fig. 4 shows circular nanoparticles, with diameters 3.65 nm and 3.56 nm, respectively. This is close to the result for
of 7.4 Æ 0.3 nm and 9.4 Æ 0.4 nm, respectively, for (1) and (2). PRWG-C18 (1) of 3.78 nm and not far from 4.22 nm for PRWG-
System (3) is characterized by an elongated anisotropic struc- (C18)2 (2) despite the presence of larger alkyl chains and
ture of the filament type with an average cross-section of 6.4 Æ different amino acid. The presence of glycine and a longer
0.5 nm and length in micrometer order. However, the system alkyl chain for (1), or two long alkyl chains (2), helps stabilize
(4) planar (bilayer) structures with 3.0 Æ 0.5 nm thickness were the hydrophobic core. This stabilization leads to smaller
found. In the contrast profile of the Cryo-TEM image for fluctuation values for the ROG when compared to PRW-O-C16
compound (3) (Fig. S6, ESI†), a region of positive and negative and PRW-NH-C16, as shown in Table S2 (ESI†). Taken as a
contrast is symmetrical to each other, where we believe to be whole, these results indicate that all spherical micellar systems
the core region and the shell of the cylindrical structures. The provide similar reaction environments for the aldol reaction to
structures and sizes obtained in these images corroborate the occur. However, (1) and (2) compounds are more stable and
models proposed in the SAXS modeling, giving greater robust- tightly packed than PRW-O-C16 and PRW-NH-C16 systems, and
ness to the results found.
that increase in packing factor translates to better catalysis.
Molecular dynamics simulation results for (1) and (2)
For simplification, systems (3) and (4) were modelled as a
revealed stable spherical micellar systems for the whole simulation cylinder and a spherical body with the same aggregation number
time in water. Our simulations show a well-defined hydrophobic as (1) and (2). As a result, a direct comparison is not possible to
core for all systems investigated in agreement with SAXS results. SAXS data. However, intrinsic quantities such as radial distribution
Radial distribution functions using the C1 alkyl carbon atom functions, area per head group, and alkyl chain volume can offer
(C1 meaning carbon 1 from the alkyl residue) as a reference for comparison data that help quantify the differences incurred by
the micelle center are shown in Fig. S9 (ESI†). We have used the adding a bulkier head group such as Lys-GCP.
backbone N, C, and O atoms to define the amino acid positions
when computing RDFs.
Radial distribution functions for systems (3) and (4) (shown
in Fig. S11, ESI†) exhibit a similar layered pattern as systems (1)
10958 | Phys. Chem. Chem. Phys., 2021, 23, 10953–10963
This journal is © the Owner Societies 2021

View Article Online
PCCP
Paper
and (2). However, these systems exhibit a prolate ellipsoid Aldol reactions
shape for the aggregation number used in the simulations
The catalytic activity of lipopeptides (1–4) in the aldol reaction
was investigated by promoting the model reaction between
p-nitrobenzaldehyde and cyclohexanone at room temperature
in water, and the results (Fig. S7 and S8, ESI†) are summarized in
Table 2. All performed reactions furnished excellent conversion
for anti aldol adduct (499%) and high levels of stereoselectivity
using lipopeptides concentrations equal to 5.0 and 2.5 mmol%
(ds 93; 7; ee 90–92%). However, a slight decrease in the stereo-
selectivity level was observed when the catalysts concentrations
decreased to 1.0 mmol%, as shown in Table 2.
We recently demonstrated that the appropriate packing of
the amphiphilic peptide assemblies can enhance stereo-
selectivity in aldol reactions.7,15,25 Furthermore, as it is known
that the water and acid additives are indispensable for the high
diastereo- and enantioselectivities in these reactions.71,72 The
excellent yields with high enantiomeric excesses obtained for
all lipopeptides (1–4), suggesting that the self-assembled
nanostructures stabilizes the reactants and intermediates in
the interface of aqueous reaction medium. The nanostructures
will also concentrate the reactants19,24 giving high levels of
conversion and stereoselectivity.
(more akin to a cylinder), resulting in broader RDF distributions.
System (3) reveals a more compact distribution compared to all
others (as expected for a strongly prolate object). Interestingly,
compounds (3) and (4) exhibit large differences in water content
distribution (as shown in RDFs in Fig. S11, ESI†). At the peak of
the proline g(r) for (3) (at r = 2.46 nm), the value for the
C1-Wat(O) g(r) is 0.62 while for (4) this value is 0.48 at r =
2.8 nm. Therefore, despite the differences in packing and head
groups, the content of water at the most probable position of
the catalytic proline is remarkably similar between systems
(1) and (3) (62% vs. 61% of the bulk value) and between systems
(2) and (4) (52% vs. 48% of the bulk value). This points to
the strong influence of the alkyl tail on determining the
environment where one expects reaction to take place. The final
configurations obtained from molecular dynamics are shown in
Fig. S13 (ESI†).
Following the work of Tanford,67 Israelachvili,68 Mitchel,
and Ninham,69 we have computed the packing parameter for
the systems (1) through (4). Alkyl chain volume calculated from
radial distribution functions obtained from molecular dynamic
simulations produced 530 Å3, 1040 Å3, 410 Å3, and 950 Å3 per
lipopeptide for systems (1) through (4), respectively. Note that
the addition of GCP increases the assembly compactness.
Assuming spherical structures, and using lc computed through
the Tanford formula (lc = 1.5 + 1.265nc, where nc is the number
of carbons in the alkyl chain), these volumes translate to areas
per lipid of 58.3 Å2, 91.4 Å2, 49.2 Å2, 86.0 Å2, and packing
parameters of 0.37, 0.46, 0.34, and 0.46, respectively. Therefore,
from conventional geometric criteria, systems (1) and (3) could
still pack as spherical objects (p E 1/3), while systems (2) and
(4) would pack either as a cylinder or bilayer shapes. However,
Connolly’s surface area70 analysis from the same trajectories
revealed much larger solvent-accessible areas for the alkyl
chains than expected. These areas were 125 Å2, 237 Å2,
140 Å2, and 267 Å2 for systems (1) through (4), respectively.
These larger areas result from the steric and electrostatic
repulsion present due to the bulky head groups that distort
the apolar surface. That increase in area per head group and
extended configurations of the alkyl chains for systems (2) (lc E
R, the core micelle radius) allowed for the size and curvature
needed for packing into a spherical object.
Molecular dynamics results indicate at least three distinct
environments in the micellar systems: the apolar core, an
intermediate hydrophilic region, and the bulk. These regions
have markedly different chemical environments and dielectric
constants and should influence L-proline protonation equilibrium.
Previous mechanistic investigations suggest that the proline
catalyst must be deprotonated for the catalysis.13,73,74 Moreover,
excess water reduces yield and enantioselectivity (see ref. 75 and
Table
2
Comparative results of the aldol reactions between p-
nitrobenzaldehyde and cyclohexanone organocatalyzed by the lipopeptides
(1–4)a
Entry Lipopeptides Catalyst (mol%) Conv.b (%) anti : sync eed (%)
1
2
3
4
5
6
7
8
(1)
(2)
(3)
(4)
(1)
(2)
(3)
(4)
(1)
(2)
(3)
(4)
5.0
5.0
5.0
5.0
2.5
2.5
2.5
2.5
1.0
1.0
1.0
1.0
499
499
499
499
499
499
499
499
499
499
499
499
93 : 7
94 : 6
93 : 7
93 : 7
93 : 7
93 : 7
93 : 7
94 : 6
92 : 8
93 : 7
93 : 7
93 : 7
91
93
91
92
90
90
90
92
78
85
86
88
The inclusion of an even bulkier head group for systems (3)
and (4) resulted in smaller apolar volumes and larger solvent-
accessible areas, indicating
a preference towards flatter
surfaces such as a cylinder, vesicle, or bilayer. This occurs
because the alkyl chain’s typical length is reduced in these
systems due to closer packing of the alkyl chains (length being
a fraction of lc) as seen during molecular dynamics and on SAXS
data. The increase in the solvent accessible area also correlates
with the more hydrophilic nature of the GCP group than
arginine, resulting in RDFs with higher water content closer to (2Â cyclohexanone volume), at room temperature, for
the hydrophobic core (see Fig. S12, ESI†). The more significant
number of hydrogen bonds allowed by the Lys-GCP group can
increase water content and closer packing.
9
10
11
12
a
Reaction
conditions:
cyclohexanone
(12
equivalent),
p-
nitrobenzaldehyde (1 equivalent), peptide (1.0–5.0 mol%) and water
days.
2
b
Conversion was determined by NMR analyses of anti aldol adduct
c
and referred to the combined yield of diastereomers. Diastereomeric
anti : syn ratios were determined by 1H NMR analysis of the crude
d
product. Determined by chiral HPLC analysis of the anti-isomer.
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10959

View Article Online
Paper
PCCP
references therein). On the other hand, Barbas and collaborators,71
inspired by aldolases, showed that efficient catalysis could happen
in water by using catalyzers that provide a more hydrophobic
environment for the transition state. This is precisely the kind of
environment provided by the micellar system.
We theorize that the micellar intermediate region can
facilitate L-proline deprotonation. Besides, TFA, present in all
experimental setups, can act as a conjugate base in a proton
shuttle mechanism, in analogy with the Houk–List
mechanism.73,74 We modelled the reaction environment using
three distinct solvents to assess the influence of the micellar
environment on the proton exchange equilibrium between
L-proline N terminal and TFA. The core, intermediate, and
exterior micelle environments were modelled as N-heptane,
cyclohexanone, and water, respectively. Of particular interest
is the micelle middle region, modelled as cyclohexanone.
Molecular dynamics indicate this is the region where most of
the proline’s reactive encounters (between cyclohexanone and
proline) should happen.
Fig. 5 Comparison of the anti-aldol conversion, for systems (1–4)
(2.5 mol%), during the time (0.5 h to 6 h), considering the reaction conditions:
cyclohexanone (12 equivalent), p-nitrobenzaldehyde (1 equivalent), and water
(2Â cyclohexanone volume), at room temperature.
presented excellent enantioselectivity and diastereoselectivity
after two days’ reaction, a condition characterized by total
aldehyde consumption.
The linear curves slope by using pseudo-first-order kinetic
velocity constant (k), considering the excess of cyclohexanone.
Computed pKa values in water using the micro solvation
approach previously described resulted in À0.56 and 10.02 for
TFA and proline (amino group), respectively. These results
compare very well to experimental ones of 0.3 and 10.6,76 with
less than one pKa unit’s errors, and validate the employed
methodology. The geometries used for the pKa calculations
are shown in Fig. S14 (ESI†).
For the (1–4) systems were found (12.1 Æ 0.3) hÀ1, (11.9 Æ 0.3) hÀ1
,
(8.6 Æ 0.4) hÀ1 and (13.6 Æ 0.2) hÀ1, respectively. These results
showed a polymorphism dependence on the aldol reaction
efficiency, in which more spherical and micellar aggregates
performed better conversions than the cylindrical shape. RDF
in the micelle’s structures observed higher water molecules
access in the hydrophobic region, justifying the reaction efficiency.
Also, the charge distribution in the molecule is an important
parameter to be considered once higher charge content at the shell
promoted more hydrophilicity comportment, as observed by SAXS,
and, consequently, better aldol efficiency was verified for the
system (4) due to the confinement of the water and the hydrophilic
portion at the materials core, with the addition of the cyclohex-
anone excess.
The proton transfer reaction between the N terminal of
L-proline and TFA is reported in Table S4 (ESI†). As expected,
the reaction is considerably endergonic in pure water, with a DG
of 14.41 kcal molÀ1. The computed value for the same reaction
in cyclohexanone shifts this value down to 4.25 kcal molÀ1
.
In this case, the micro-solvation approach is closer to the micro-
heterogeneous environment present in the intermediate region
of the micelles, where a substantial amount of water is still
available. Still, one expects a lower dielectric medium when
compared to bulk water. In n-heptane, the proton
transfer reaction becomes strongly exergonic, with a DG of
À41.12 kcal molÀ1. Therefore, the closed to the micelle center,
the easier it is for the proton transfer reaction to occur.
Our results indicate that in regions of the micelle that are
proline-rich (the intermediate realm, modelled as cyclohexanone),
the proton transfer reaction is strongly enhanced compared to
bulk solvent. Thus, we conclude that the micellar environment is
critical in facilitating the initial steps of enamine formation in
water by enabling the proton abstraction from proline by TFA
conjugated base. We reason that, under micellar conditions, TFA
possibly participates in a proton shuttle mechanism during
catalysis.
It is observed that lipopeptides with one aliphatic chain
(micellar morphology), (1), and cylindrical, (3) initially present the
best conversion rates (from 0.5 h to 1 h). During the reaction (2 h),
lipopeptide (4) (which has populations of both spherical micelles
and bilayers) gains prominence over the others. At the end of 6 h
of reaction, it is together with micellar lipopeptide (1), presenting
the highest conversions. Therefore, cylindrical morphology from
(3) favors only the beginning of the reaction (up to 1 h). After that
time, there is a less pronounced increase in conversion, probably
due to the packing factor. In cylindrical morphology, the packing
is denser, providing a slower diffusion of the aldehyde within the
confined system. On the other hand, for spherical micelles’
morphology, diffusion is easier, leading to higher conversion
rates over the reaction time. As shown in Table S1 of the ESI,†
in 48 h, all reactions with the different lipopeptides have already
reached a steady state with maximum conversion.
Kinetics experiments were also performed to understand the
dependence of the anti-aldol product conversion varying the
time (0.5–6 h). Fig. 5 shows that, initially, better conversion
results were observed for the one lipid chain lipopeptides,
reaching 33% and 30% for (1) and (3), respectively. During that
time, an enhancement of the sample (4) conversion was
Conclusions
observed, which obtained almost 100% after 6 h reaction, In this work, we studied the self-assembly process and the
together with (1). However, as seen by Table 2, all samples effect of the polymorphism of lipopeptides containing the PRW
10960
|
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963
This journal is © the Owner Societies 2021

View Article Online
PCCP
Paper
tripeptide attached to one or two lipid chains and in the Acknowledgements
absence or presence of the GCP group. These materials
presented an enhancement of the cac from 3.4 Â 10À4 wt% to
2.5 Â 10À3 wt%, for the sequences with the GCP, suggesting a
higher, stronger packing of the material. The systems were
characterized by b-sheet secondary structures. By Cryo-TEM
and SAXS were observed spherical nanostructures for systems
(1), (2), while system (3) presented cylindrical shape and (4)
bilayer structure, indicating the presence of GCP group
chances drastically the self-assembly of the systems. We saw
that all systems act as excellent catalysts for the aldol
reactions by catalytic assays, using p-nitrobenzaldehyde and
cyclohexanone. However, the change of the hydrophobicity of
the system by the lipid chain and/or by the GCP group
promoted an enhancement of the packing factor for com-
pounds (3) and (4) as shown by SAXS data and molecular
dynamics simulations.
This work was supported by FAPESP (grant no. 2017/02317-2) and
CNPq (grant no. 304389/2019-6). INCT in Bioanalytics (FAPESP
grant no. 2014/50867-3 and CNPq grant no. 465389/2014-7) is
kindly acknowledged for the grants. J. N. B. D. P. and B. B. G.
acknowledge FAPESP (projects number 2015/20446-9 and 2018/
05888-3) for a doctoral and post-doctoral fellowship, respectively.
The staff at LNNano and LNLS are kindly recognized for invalu-
able help and access to TEM (project number TEM-23340) and
SAXS Beamline (SAXS1-20180103) facilities. The authors are grate-
ful to the Multiuser Central Facilities at UFABC and UNIFESP.
References
1 R. Mammadov, A. B. Tekinay, A. Dana and M. O. Guler,
Micron, 2012, 43, 69–84.
Results from molecular dynamics simulations paint an
atomistic view of the internal micellar structure that offers
explanations over the influence of different alkyl chains and
amino acids on their network. Radial distribution functions
revealed a stable layered structural organization of the residues
to the micellar center for all systems. The presence of GCP lead
to more tightly packed structures in the models investigated.
Moreover, differences in the water content observed close to the
proline position correlate with larger initial (0.5 and 1 h) anti-
aldol conversion percentages. Based on Density Functional
Calculations, we propose that the micellar environment is
critical in facilitating the initial steps in enamine formation
by enabling the proton abstraction from proline by TFA.
We reason that, under micellar conditions, TFA possibly
participates in a proton shuttle mechanism in enamine
formation. It is essential to mention that other micellar effects,
such as controlling reagent and product diffusion, can be at
2 A. Dehsorkhi, V. Castelletto and I. W. Hamley, J. Pept. Sci.,
2014, 20, 453–467.
3 E. R. Silva, E. Listik, S. W. Han, W. A. Alves, B. M. Soares,
M. Reza, J. Ruokolainen and I. W. Hamley, Biophys. Chem.,
2018, 233, 1–12.
4 I. W. Hamley, Soft Matter, 2011, 7, 4122–4138.
5 J. Pelin, C. J. C. Edwards-Gayle, V. Castelletto, A. M. Aguilar,
W. A. Alves, J. Seitsonen, J. Ruokolainen and I. W. Hamley,
ACS Appl. Mater. Interfaces, 2020, 12, 13671–13679.
6 V. Castelletto, C. J. C. Edwards-Gayle, I. W. Hamley,
J. N. B. D. Pelin, W. A. Alves, A. M. Aguilar, J. Seitsonen
and J. Ruokolainen, ACS Appl. Bio Mater., 2019, 2, 3639–3647.
7 J. N. B. D. Pelin, B. B. Gerbelli, B. M. Soares, A. M. Aguilar
and W. A. Alves, Catal. Sci. Technol., 2019, 9, 4304–4313.
8 G. Bredig and P. Fiske, Biochem. Z., 1912, 46, 7–23.
9 Y. Yamashita, T. Yasukawa, W.-J. Yoo, T. Kitanosono and
S. Kobayashi, Chem. Soc. Rev., 2018, 47, 4388–4480.
play, resulting in the observed differences in enantiomeric 10 Z. G. Hajos and D. R. Parrish, J. Org. Chem., 1974, 39,
excess.
1615–1621.
11 U. Eder, G. Sauer and R. Weichert, Angew. Chem., Int. Ed.
Engl., 1971, 10, 496–497.
12 C. Pidathala, L. Hoang, N. Vignola and B. List, Angew.
Chem., Int. Ed., 2003, 42, 2785–2788.
Author contributions
Conceptualization, W. A. A., A. M. A., C. S. and M. D. C. N.;
lipopeptide synthesis and aldol reactions, B. M. S. and A. M. A.;
fluorescence measurements, K. B. A. and J. N. B. D. P.; cryo-
TEM, M. A. F. and R. V. P.; molecular dynamics simulations and
quantum mechanical calculations, P. T. S. and M. D. C. N.;
SAXS experiments, E. R. S., B. B. G. and W. A. A.; formal
analysis, W. A. A., B. M. S., A. M. A., E. R. S., B. B. G. and
M. D. C. N.; writing—original draft preparation, B. M. S. and
B. B. G.; writing—review and editing, W. A. A., A. M. A., E. R. S.
and M. D. C. N.; supervision, W. A. A.; funding acquisition,
W. A. A. All authors have read and agreed to the published
version of the manuscript.
13 B. List, R. A. Lerner and C. F. Barbas, J. Am. Chem. Soc., 2000,
122, 2395–2396.
14 K. A. Ahrendt, C. J. Borths and D. W. C. MacMillan, J. Am.
Chem. Soc., 2000, 122, 4243–4244.
15 A. M. Aguilar, M. Bruna, J. N. B. D. Pelin, B. B. Gerbelli and
W. A. Alves, in Peptide-based Biomaterials, ed. M. O. Guler,
The Royal Society of Chemistry, 2021, ch. 5, pp. 126–173.
16 A. J. Metrano, A. J. Chinn, C. R. Shugrue, E. A. Stone, B. Kim
and S. J. Miller, Chem. Rev., 2020, 120, 11479–11615.
´
17 F. Rodrıguez-Llansola, J. F. Miravet and B. Escuder, Chem.
Commun., 2009, 7303–7305.
´
18 F. Rodrıguez-Llansola, B. Escuder and J. F. Miravet, J. Am.
Chem. Soc., 2009, 131, 11478–11484.
19 B. M. Soares, A. M. Aguilar, E. R. Silva, M. D. Coutinho-Neto,
I. W. Hamley, M. Reza, J. Ruokolainen and W. A. Alves, Phys.
Chem. Chem. Phys., 2017, 19, 1181–1189.
Conflicts of interest
The authors declare no competing financial interests.
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10961

View Article Online
Paper
PCCP
20 V. Castelletto, C. J. C. Edwards-Gayle, I. W. Hamley, 40 A. W. Gotz, M. J. Williamson, D. Xu, D. Poole, S. Le Grand
J. N. B. D. Pelin, W. A. Alves, A. M. Aguilar, J. Seitsonen
and J. Ruokolainen, ACS Appl. Bio Mater., 2019, 2,
3639–3647.
and R. C. Walker, J. Chem. Theory Comput., 2012, 8,
1542–1555.
41 R. Salomon-Ferrer, D. A. Case and R. C. Walker, Wiley
Interdiscip. Rev.: Comput. Mol. Sci., 2013, 3, 198–210.
42 H. J. C. Berendsen, J. P. M. Postma, W. F. Vangunsteren,
A. Dinola and J. R. Haak, J. Chem. Phys., 1984, 81, 3684–3690.
43 R. J. Loncharich, B. R. Brooks and R. W. Pastor, Biopolymers,
1992, 32, 523–535.
44 J. P. Ryckaert, G. Ciccotti and H. J. C. Berendsen, J. Comput.
Phys., 1977, 23, 327–341.
45 D. R. Roe and T. E. Cheatham, J. Chem. Theory Comput.,
2013, 9, 3084–3095.
21 M. Z. Shao, Q. X. Jin, L. Zhang and M. H. Liu, Chem. Lett.,
2012, 41, 1349–1350.
22 O. Zozulia, M. A. Dolan and I. V. Korendovych, Chem. Soc.
Rev., 2018, 47, 3621–3639.
23 P. R. Arivalagan and Y. Zhao, Org. Biomol. Chem., 2015, 13,
770–775.
24 N. Mase, N. Noshiro, A. Mokuya and K. Takabe, Adv. Synth.
Catal., 2009, 351, 2791–2796.
25 J. N. B. D. Pelin, C. J. C. Edwards-Gayle, A. M. Aguilar, A. Kaur,
I. W. Hamley and W. A. Alves, Soft Matter, 2020, 16, 4615–4624. 46 W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics
26 C. Schmuck, Chem. – Eur. J., 2000, 6, 709–718. Modell., 1996, 14, 33–38.
27 M. Li, M. R. Stojkovic, M. Ehlers, E. Zellermann, 47 E. Rossini and E. W. Knapp, J. Comput. Chem., 2016, 37,
I. Piantanida and C. Schmuck, Angew. Chem., Int. Ed.,
2016, 55, 13015–13018.
28 A. P. Hammersley, J. Appl. Crystallogr., 2016, 49, 646–652.
29 I. Bressler, J. Kohlbrecher and A. F. Thu¨nemann, J. Appl.
Crystallogr., 2015, 48, 1587–1598.
30 A. A. Skjevik, B. D. Madej, R. C. Walker and K. Teigen,
J. Phys. Chem. C, 2012, 116, 11124–11136.
31 C. J. Dickson, B. D. Madej, A. A. Skjevik, R. M. Betz,
2163–2164.
48 Z. Szakacs and B. Noszal, J. Math. Chem., 1999, 26, 139–155.
49 F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005,
7, 3297–3305.
50 J. J. Zheng, X. F. Xu and D. G. Truhlar, Theor. Chem. Acc.,
2011, 128, 295–305.
51 P. J. Stephens, F. J. Devlin, C. F. Chabalowski and
M. J. Frisch, J. Phys. Chem., 1994, 98, 11623–11627.
K. Teigen, I. R. Gould and R. C. Walker, J. Chem. Theory 52 A. D. Becke, J. Phys. Chem., 1993, 98, 5648–5652.
Comput., 2014, 10, 865–879.
32 D. A. Case, I. Y. Ben-Shalom, S. R. Brozell, D. S. Cerutti,
53 C. T. Lee, W. T. Yang and R. G. Parr, Phys. Rev. B: Condens.
Matter Mater. Phys., 1988, 37, 785–789.
T. E. Cheatham, III, V. W. D. Cruzeiro, T. A. Darden, 54 S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Phys. Chem.,
R. E. Duke, D. Ghoreishi, M. K. Gilson, H. Gohlke, 2010, 132, 154104.
A. W. Goetz, D. Greene, R. Harris, N. Homeyer, Y. Huang, 55 S. Grimme, S. Ehrlich and L. Goerigk, J. Chem. Comput.,
S. Izadi, A. Kovalenko, T. Kurtzman, T. S. Lee, S. LeGrand, 2011, 32, 1456–1465.
P. Li, C. Lin, J. Liu, T. Luchko, R. Luo, D. J. Mermelstein, 56 F. Neese, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2018,
K. M. Merz, Y. Miao, G. Monard, C. Nguyen, H. Nguyen, 8, e1327.
I. Omelyan, A. Onufriev, F. Pan, R. Qi, D. R. Roe, A. Roitberg, 57 A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys.
C. Sagui, S. Schott-Verdugo, J. Shen, C. L. Simmerling, Chem. B, 2009, 113, 6378–6396.
J. Smith, R. Salomon-Ferrer, J. Swails, R. C. Walker, 58 J. R. Pliego and J. M. Riveros, J. Phys. Chem. A, 2002, 106,
J. Wang, H. Wei, R. M. Wolf, X. Wu, L. Xiao, D. M. York 7434–7439.
and P. A. Kollman, AMBER 2018, University of California, 59 R. M. P. Silva, G. D. Silva, M. D. Coutinho-Neto and
San Francisco, 2018. H. B. Suffredini, Electrochim. Acta, 2016, 212, 195–200.
33 J. A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, 60 D. W. Pierce and S. G. Boxer, Biophys. J., 1995, 68, 1583–1591.
K. E. Hauser and C. Simmerling, J. Chem. Theory Comput., 61 R. M. Clegg, Curr. Opin. Biotechnol., 1995, 6, 103–110.
2015, 11, 3696–3713.
34 J. M. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman and 63 D. I. Svergun and M. H. J. Koch, Rep. Prog. Phys., 2003, 66,
D. A. Case, J. Comput. Chem., 2004, 25, 1157–1174. 1735–1782.
35 C. I. Bayly, P. Cieplak, W. D. Cornell and P. A. Kollman, 64 G. Pabst, M. Rappolt, H. Amenitsch and P. Laggner, Phys.
62 J. S. Pedersen, Adv. Colloid Interface Sci., 1997, 70, 171–210.
J. Phys. Chem., 1993, 97, 10269–10280.
36 D. A. Case, T. E. Cheatham, T. Darden, H. Gohlke, R. Luo,
Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top.,
2000, 62, 4000–4009.
K. M. Merz, A. Onufriev, C. Simmerling, B. Wang and 65 V. Castelletto, R. M. Gouveia, C. J. Connon and
R. J. Woods, J. Comput. Chem., 2005, 26, 1668–1688. I. W. Hamley, Faraday Discuss., 2013, 166, 381–397.
37 W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, 66 V. Castelletto and I. W. Hamley, Methods Mol. Biol., 2018,
R. W. Impey and M. L. Klein, J. Chem. Phys., 1983, 79,
1777, 3–21.
926–935.
67 C. Tanford, J. Phys. Chem., 1972, 76, 3020–3024.
38 L. Martinez, R. Andrade, E. G. Birgin and J. M. Martinez, 68 J. Israelachvli, Biochim. Biophys. Acta, 1977, 469, 221–225.
J. Comput. Chem., 2009, 30, 2157–2164.
39 J. M. Martinez and L. Martinez, J. Comput. Chem., 2003, 24,
819–825.
69 D. J. Mitchell and B. W. Ninham, J. Chem. Soc., Faraday
Trans., 1981, 77, 601–629.
70 M. L. Connolly, J. Appl. Crystallogr., 1983, 16, 548–558.
10962 | Phys. Chem. Chem. Phys., 2021, 23, 10953–10963
This journal is © the Owner Societies 2021

View Article Online
PCCP
Paper
71 N. Mase, Y. Nakai, N. Ohara, H. Yoda, K. Takabe, F. Tanaka 74 A. Armstrong, R. A. Boto, P. Dingwall, J. Contreras-Garcia,
and C. F. Barbas, J. Am. Chem. Soc., 2006, 128, 734–735.
72 Y. Hayashi, T. Sumiya, J. Takahashi, H. Gotoh, T. Urushima
and M. Shoji, Angew. Chem., Int. Ed., 2006, 45, 958–961.
73 F. R. Clemente and K. N. Houk, Angew. Chem., Int. Ed., 2004,
43, 5766–5768.
M. J. Harvey, N. J. Mason and H. S. Rzepa, Chem. Sci., 2014,
5, 2057–2071.
75 N. Mase and C. F. Barbas, Org. Biomol. Chem., 2010, 8,
4043–4050.
76 G. C. Fonger, Toxicology, 1995, 103, 137–145.
This journal is © the Owner Societies 2021
Phys. Chem. Chem. Phys., 2021, 23, 10953–10963 | 10963