Enantioseparation of mandelic acid and substituted derivatives by high-performance liquid chromatography with hydroxypropyl-β-cyclodextrin as chiral mobile additive and evaluation of inclusion complexes by molecular dynamics

Article abstract of DOI:10.1002/chir.23348

The enantioseparation and resolution mechanism of mandelic acid (MA), 4-methoxymandelic acid (MMA), and 4-propoxymandelic acid (PMA) were investigated by reversed-phase high-performance liquid chromatography (HPLC) with 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) as a chiral mobile-phase additive and molecular dynamics simulation. The suitable chromatographic conditions for the enantioseparation of MA, MMA, and PMA were obtained. Under the selected chromatographic conditions, these enantiomers could achieve baseline separation. The results of thermodynamic parameter analysis revealed that the main driven forces for the enantioseparation of MA, MMA, and PMA could be van der Waals forces and hydrogen-bonding interactions and the chromatographic retention of these chiral compounds was an enthalpy-driven process. The results of the molecular simulation revealed that their chiral resolution mechanism on HP-β-CD was responsible for the formation of inclusion complexes of enantiomers with HP-β-CD with different conformations and binding energies. And the binding energy of HP-β-CD with (S)-isomer was larger than that with (R)-isomer, which is consistent with the experimental results of the first elution of (S)-isomer. Additionally, it is also confirmed that the interaction energies included the van der Waals energy (?E_vdw), electrostatic energy (?E_elec), polar solvation energy, and SASA energy (?E_sasa), and the separation factor (α) was closely connected with the disparity in the binding energies of optical isomers and HP-β-CD complexes. Meanwhile, from molecular dynamics simulation, it can be found that the ?(?E_binding), (?(?E_binding) = ?E_binding,R ? ?E_binding,S) value was in order of MA–HP-β-CD complex > MMA–HP-β-CD complex > PMA–HP-β-CD complex, which was consistent with the order of Δ(ΔG) values obtained from van't Hoff plot. This indicated that the molecular dynamics simulation has predictive function for chiral resolution.

Full text of DOI:10.1002/chir.23348

Received: 21 February 2021
DOI: 10.1002/chir.23348
Revised: 15 July 2021
Accepted: 21 July 2021
R E S E A R C H A R T I C L E
Enantioseparation of mandelic acid and substituted
derivatives by high-performance liquid chromatography
with hydroxypropyl-β-cyclodextrin as chiral mobile additive
and evaluation of inclusion complexes by molecular
dynamics
Jie-Hua Shi
| Zhen-Yi Lin | Song-Bo Kou | Bao-Li Wang | Shao-Liang Jiang
College of Pharmaceutical Science,
Zhejiang University of Technology,
Hangzhou, P.R. China
Abstract
The enantioseparation and resolution mechanism of mandelic acid (MA),
4-methoxymandelic acid (MMA), and 4-propoxymandelic acid (PMA) were
investigated by reversed-phase high-performance liquid chromatography
(HPLC) with 2-hydroxypropyl-β-cyclodextrin (HP-β-CD) as a chiral mobile-
phase additive and molecular dynamics simulation. The suitable chromato-
graphic conditions for the enantioseparation of MA, MMA, and PMA were
obtained. Under the selected chromatographic conditions, these enantiomers
could achieve baseline separation. The results of thermodynamic parameter
analysis revealed that the main driven forces for the enantioseparation of MA,
MMA, and PMA could be van der Waals forces and hydrogen-bonding interac-
tions and the chromatographic retention of these chiral compounds was an
enthalpy-driven process. The results of the molecular simulation revealed that
their chiral resolution mechanism on HP-β-CD was responsible for the forma-
tion of inclusion complexes of enantiomers with HP-β-CD with different con-
formations and binding energies. And the binding energy of HP-β-CD with
(S)-isomer was larger than that with (R)-isomer, which is consistent with the
experimental results of the first elution of (S)-isomer. Additionally, it is also
confirmed that the interaction energies included the van der Waals energy
(ΔEvdw), electrostatic energy (ΔEelec), polar solvation energy, and SASA energy
(ΔEsasa), and the separation factor (α) was closely connected with the disparity
in the binding energies of optical isomers and HP-β-CD complexes. Mean-
while, from molecular dynamics simulation, it can be found that the Δ
(ΔEbinding), (Δ(ΔEbinding) = ΔEbinding,R ꢀ ΔEbinding,S) value was in order of
MA–HP-β-CD complex > MMA–HP-β-CD complex > PMA–HP-β-CD com-
plex, which was consistent with the order of Δ(ΔG) values obtained from van't
Hoff plot. This indicated that the molecular dynamics simulation has predic-
tive function for chiral resolution.
Correspondence
Jie-Hua Shi and Shao-Liang Jiang, College
of Pharmaceutical Science, Zhejiang
University of Technology, 18, Chaowang
Road, Hangzhou, P.R. China.
Email: shijh@zjut.edu.cn;
shljiang@zjut.edu.cn
Funding information
Huahai Pharmaceutical Co., Ltd., Grant/
Award Number: YX-KF-[2005]018
Chirality. 2021;1–10.
wileyonlinelibrary.com/journal/chir
© 2021 Wiley Periodicals LLC.
1

2
SHI ET AL.
K E Y W O R D S
4-methoxymandelic acid, 4-propoxymandelic acid, chiral recognition, cyclodextrin, HPLC,
mandelic acid, molecular dynamics simulation
1 | INTRODUCTION
mechanism based on molecular docking.18 Shi et al. stud-
ied the chiral separation of optical isomers of methyl
Since the US Food and Drug Administration (FDA)
announced the guidelines for chiral drug in 1992, enan-
tiomeric separation of chiral compounds has received
much attention. Up to now, high-performance liquid
chromatography (HPLC) has become an essential tool
for enantiomer separation of chiral compounds in the
pharmaceutical, agrochemical, perfume, textile, or food
industries.1 As is known to all, cyclodextrins, a kind of
natural cyclic oligosaccharides, possess much better chi-
ral recognition ability for many chiral compounds.2,3
This chiral recognition ability is responsible for which
they can be easy-to-generate complexes with diverse
guest molecules and thus change the nature of guest
molecules.4 2-Hydroxypropyl-β-cyclodextrin (HP-β-CD),
one of the β-CD derivatives, possesses better chiral rec-
ognition ability and solubility in aqueous solution than
native β-cyclodextrin (β-CD).5 Recently, HP-β-CD
has been widely used as chiral selector to separate
the enantiomers of chiral compounds, such as
using HPLC,6,7 capillary electrophoresis (CE),8,9
enantioselective liquid–liquid extraction,10 and mem-
brane separation.11
As the important chiral building blocks and organic
synthesis intermediates, mandelic acid (MA) and its
derivatives are extensively utilized for pharmaceutical
and fine chemical industries. For example, racemic MA
was used for the biosynthesis of L-phenylglycine.12
2-Substituted-1,4-diketones were prepared from (S)-MA
enolate through enantioselective synthesis.13 (R)-MA was
used as an antibiotic side-chain modifier for
cefidiomycetes, and (S)-MA is a precursor for the synthe-
sis of the drug (S)-oxibutin for the treatment of urinal
urgency, frequent urination, and urinary incontinence.14
mandelate and methyl cyclohexylmandelate using chiral
stationary phase method and explored the chiral recogni-
tion mechanism using quantum chemical calculation.19
Zhang et al. studied the chiral separation of
α-cyclopentyl-mandelic acid enantiomers by continuous
liquid–liquid extraction.20 In short, the use of a chiral
mobile-phase additive to perform the enantioseparations
of MA and its derivatives is feasible. Recently, theoretical
methods such as semiempirical methods, molecular
mechanics (MM), and molecular dynamics (MD) are
widely used to study the inclusion complexes of the CDs
or their derivatives with guests and to comprehend the
chiral resolution mechanism.18,19,21–23
The aim of this research was to explore the chiral sep-
aration of MA, MMA, and PMA enantiomers (Figure 1)
by HPLC with HP-β-CD as a chiral mobile-phase additive
and to comprehend the chiral resolution mechanism at
the molecular level through molecular simulations
including molecular docking and MD.
2 | MATERIALS AND METHODS
2.1 | Chemicals and solutions
MA (≥99%) was purchased from Aladdin Chemistry Co.,
Ltd. (Shanghai, China). PMA (≥98%) was purchased from
Adamas-Beta Reagent, Ltd. (Shanghai, China). MMA
(≥97%) was purchased from Macklin Chemistry Co., Ltd.
(Shanghai, China). HP-β-CD (MS = 6.4) was provided by
DELI Biochemical Industry Co., Ltd. (Xi'an, China).
Phosphoric acid was obtained from Lingfeng Chemical
Reagent Co., Ltd. (Shanghai, China). Sodium dihydrogen
phosphate was purchased from Guanghua Sci-Tech Co.,
Ltd. (Guangdong, China). Acetonitrile and methanol
used for HPLC analysis were of chromatographic grade.
Other reagents were of analytical grade and directly used
without further purification.
Naphthodifuranone
dyes
were
prepared
from
4-alkoxymandelic acid such as 4-propoxymandelic acid
(PMA) and 4-methoxymandelic acid (MMA).15 At the
same time, many studies on the chiral separation of MA
and its derivatives were performed. Tong et al. reported
that the optical isomers of MA derivatives were separated
using chiral mobile-phase additive method.16 Tong et al.
also studied the enantioseparation of MA using counter-
current chromatography.17 Shi et al. investigated the sep-
aration of optical isomers of α-cyclohexylmandelic acid
and methyl α-cyclohexylmandelates with HP-β-CD as
chiral selector and discussed their chiral recognition
The HP-β-CD was dissolved in phosphate-buffered
solution (PBS) (0.1 mol lꢀ1
)
containing HP-β-CD
(15 mmol lꢀ1) and adjusted to the required pH value
using phosphoric acid, which solution was used as one of
the components of mobile phase and filtered using
microfiltration membrane (0.45 μm) before use. All test-
ing solutions of MA, MMA, and PMA were prepared in

SHI ET AL.
3
FIGURE 1 Structures of MA, MMA, PMA, and HP-
β-CD
anhydrous methanol with the concentration of about
Gaussian 03 until no imaginary frequency and the opti-
mized structure of HP-β-CD was represented in Figure 1.
(R/S)-MA and (R/S)-MMA were downloaded from
the PubChem. (R/S)-PMA was constructed with the help
of Gaussian view by substituting para of benzene ring of
(R/S)-MA with propoxy group. These enantiomeric struc-
tures were optimized using DFT/B3lyp/6-31+g(d,p)
approach until all eigenvalues of the Hessian matrix were
positive.
The inclusion complex of HP-β-CD with various
enantiomers was first simulated by semiflexible docking
technique with the help of AutoDock program (Version
4.2). During molecular docking, the HP-β-CD was set as a
rigid molecule, whereas guest molecules were allowed to
move in free motion. A grid map with the grid points of
50 ꢁ 50 ꢁ 40 with grid spacing 0.375 Å was set to ensure
that HP-β-CD is completely surrounded. Lamarckian
genetic algorithm approach was used for searching the
structure of inclusion complex. The running times and
the number of evaluations were fixed at 500 and
2,500,000.23 Other docking parameters were set as
default. In this work, all parameters were the same for
each docking. The structures of all inclusion complexes
with lowest acting energy were used for the starting
structure of MD simulation.
2 mmol lꢀ1
.
2.2 | HPLC analysis
All chromatographic separations were carried out on
Waters 2695 Separations Module HPLC equipped with
Waters 2996 Photodiode Array Detector and Empower
2 Chemstation workstation (Waters Corporation, Shang-
hai, China). XBridge™ C18 (250 ꢁ 4.6-mm i.d., 5 μm)
column was used in whole research. The mixture solu-
tion of acetonitrile and 0.1-mol lꢀ1 PBS containing
15-mmol lꢀ1 HP-β-CD was used as mobile phase. The
detected wavelength and column temperature were fixed
at 230 nm and 30ꢂC, respectively. All the testing solutions
of 10 μl were injected into the HPLC for analysis.
2.3 | MD simulation
In this work, the used HP-β-CD was derived from the
selective condensation reaction in position 6 of β-CD, and
the mean substitution was 6.4. Therefore, we supposed
the structure of HP-β-CD was mainly hepta-(6-O-
(2-hydropropyl))-β-CD by the reference of the reported
literature,21,22 which was used in theoretical calculation.
The supposed structure of HP-β-CD was built from the
coordinate of β-CD, which was extracted from the crystal
structure (PDB: 1DMB) through the replacement of the
H-atoms in all 6-OH with 2-hydroxypropyl group. After
energetic and structural filtering to remove high-energy
and redundant conformers, the resulting structure was
optimized by B3LYP/6-31G(d,p) method implemented in
All MD simulations were performed on GROMACS
program (Version 2019.3). First, the restrained electro-
static potential (RESP) of each guest molecule was calcu-
lated at DFT/B3lyp/6-311G** level. Then, each isomer
with the RESP charges was parameterized by GAFF force
field.24 Other force field parameters came from the pro-
posed parameters of ANTECHAMBER.25,26 Each inclu-
sion complex with stoichiometric ratio of 1:1 was
solvated in a regular 1.6-nm dodecahedron box with

4
SHI ET AL.
TIP3P water containing 5% acetonitrile (v/v) model, to
ensure that the whole system was dissolved in
acetonitrile–water system. Next, the system was treated
by energy minimization using a steepest descent integra-
tor (50,000 steps) until the maximum force of
<10.0 kJ molꢀ1 nmꢀ1 to remove the largest strains. After
that, the whole system was gradually warmed to 300 K
under the NVT ensemble, followed by the NPT ensemble
at 1 bar. Finally, MD simulation was performed for
120 nsec in 60,000,000-step production. The representa-
tive geometry of each inclusion complex was obtained by
the cluster analysis using the “gmx cluster” program. The
number of collected frames was 12,000, and the balanced
trajectories from 20 to 100 nsec were extracted for further
analysis of interaction forces.
additive system, (S)-isomer was first eluted, suggesting
that the interaction of HP-β-CD with (S)-isomer was
stronger than that with (R)-isomer.
Subsequently, the effect of the concentration of HP-
β-CD additive on the chiral separation was investigated,
and the results were summarized in Table S1. The results
revealed that the capacity factors (k) decreased with the
increase of the concentration of HP-β-CD whereas
the separation factor (α) increased, indicating that the
elution capacity is enhanced due to the solubilization
effect of HP-β-CD and the separation selectivity
increased. However, the resolution factor (Rs) of these
enantiomers gradually declined with the increase of HP-
β-CD concentration. So, considering the retention and
resolution capabilities of these chiral compounds, the
concentration of HP-β-CD additive (15 mmol lꢀ1) was
used for the enantioseparation of these chiral
compounds.
3 | RESULTS AND DISCUSSION
3.1 | Assessment on chromatographic
conditions
3.1.2 | Influence of pH value
3.1.1 | Influence of the type and
concentration of chiral mobile-phase additive
As is known to all, MA, MMA, and PMA belong to strong
organic acid.16 Therefore, an increase in the pH of the
aqueous solution will result in a decrease in the content
of free MA derivative enantiomers (electrically neutral).
In addition, the chiral recognition ability of HP-β-CD is
mainly related to the affinity of the neutral isomer rather
than ionic isomer.27 This means that the pH value of the
mobile-phase affects obviously on the enantiomeric sepa-
ration of acidic chiral compounds. In this work, the effect
of pH values in the mobile phase containing HP-β-CD on
the retention behaviors and enantioselectivity of MA,
MMA, and PMA was determined, and the findings were
summarized in Table S2. The finding revealed that, with
the increases of pH value, the capacity factors (k1 and k2),
separation factor (α), and resolution factor (Rs) of these
enantiomers gradually decreased, which was inconsistent
In our exploratory experiments, β-CD, HP-β-CD, and
sulfobutylether-β-CD (SBE-β-CD) were used as chiral
mobile-phase additives, respectively, for the enantiomeric
separation of MA, MMA, and PMA. The results revealed
that their optical isomers can be efficiently resolved when
HP-β-CD is used as chiral mobile-phase additives and the
(S)-isomer was eluted before (R)-isomer (Figure 2). These
optical isomers cannot be efficiently separated using
SBE-β-CD and β-CD as chiral mobile-phase additives,
although β-CD has certain chiral recognition ability to
PMA. This indicates that HP-β-CD has a good chiral rec-
ognition ability for the separation of the optical isomers
of MA, MMA, and PMA. In the HP-β-CD mobile-phase
FIGURE 2 Chromatograms of the
separations of (A) MA, (B) MMA, and (C) PMA
enantiomers when β-CD, HP-β-CD, and SBE-
β-CD as chiral mobile-phase additives.
Chromatographic conditions: mobile phase:
acetonitrile:PBS (0.1 mol lꢀ1, pH 2.50)
containing HP-β-CD (10 mmol lꢀ1) = 5:95
(v/v). Flow rate: 0.6 ml minꢀ1; detection
wavelength: 230 nm; column temperature: 30ꢂC

SHI ET AL.
5
with the result reported by Tong et al.16 However, when
the pH value was 3.0, the three chiral compounds have
suitable retention capacity and can achieve baseline sepa-
ration. So, PBS with pH 3.0 was used for the
enantioselectivity of MA, MMA, and PMA.
MMA, and PMA was further investigated, and the
findings were listed in Table S4. The findings revealed
that the k and α values declined gradually with rising
column temperature, which is responsible for the
decline in the partition coefficient of solutes between
stationary phase and mobile phase. However, with ris-
ing temperature, the Rs values of MA and MMA
declined whereas the Rs value of PMA increased. To
our knowledge, the resolution factor is proportional to
the square root of theoretical plate number (N),
(α ꢀ 1)/α, and proportional to k/(1 + k). Obviously, the
decline in the α values with rising temperature is not
good for improving the resolution factor. Nevertheless,
the rising temperature can increase the diffusion coeffi-
cient of solute in the mobile phase and stationary
phase and thus accelerates the partitioning between
the two phases, resulting in the increase in the theoret-
ical plate number and in favor of the enhancement in
the resolution factor (Rs).31 This indicates that the
influence of column temperature on the resolution fac-
tor (Rs) of these chiral compounds is quite compli-
cated, resulting in the increase in the Rs value for
PMA with higher k value and the decrease in the Rs
value for MA and MMA with smaller k value. This
also means that the kinetic factors (theoretical plate
number) play a leading role in the effect of column
temperature on the Rs value for PMA whereas the
thermodynamic factors (such as k and α values) play a
leading role in the effect of column temperature on
the Rs value for MA and MMA.
3.1.3 | Influence of proportion of acetonitrile
The different hydro-organic mobile phases made up with
water and acetonitrile were tested, and the results were
summarized in Table S3. The results indicated that the
values of k and α of these enantiomers gradually
increased with a decrease of acetonitrile content in
mobile phase containing HP-β-CD, indicating that the
proportion of acetonitrile in the mobile phase obviously
affected the partition coefficient of each enantiomer
between stationary phase (C18 column) and mobile
phase28,29 and thus affected the k value of each enantio-
mer. Combined with the influence of HP-β-CD on the
chromatographic retention, it can be inferred that
the retention behavior of each enantiomer is driven by its
partition coefficient between the mobile and stationary
phases and its complexation with cyclodextrins.30 This
also means that the enantioselectivity is dominated by
the distribution of chiral compounds between the mobile
and stationary phases and the inclusion interaction
between chiral compound and cyclodextrins in the chro-
matographic system with chiral mobile-phase additives.
From Table S3, it can be also found that the column effi-
ciency (theoretical plate number) increases with the
increase of acetonitrile content in mobile phase.
Although the enhancement in the column efficiency is
beneficial to the improvement of the resolution factor
(Rs), the decline in the α value is not conducive to
improving the enantioselectivity. Finally, the Rs values of
these enantiomers gradually decrease with the increase
of acetonitrile content in mobile phase.
Meanwhile, considering the resolution factor and
analytic time as well as column life, the selected col-
umn temperature of 30ꢂC is suitable for simultaneous
analysis of MA, MMA, and PMA.
3.1.5 | Influence of flow rate
Additionally, it can be found that when the propor-
tion of acetonitrile in the mobile phase was more than
9% (v/v), the resolution (Rs) of MA, and MMA isomers
was less than 1.5, could not achieve baseline separation.
In terms of economy, environmental protection, separa-
tion effect, and retention time, the optimal mobile phase
is the ratio of acetonitrile and phosphate buffer
(0.1 mol lꢀ1, pH 3.0) containing 15-mmol lꢀ1 HP-β-CD is
5:95 (v/v).
The effect of flow rate on the enantioselectivity of the
selected MA and its derivatives was investigated using
PBS (0.1 mol lꢀ1, pH 3.00) containing 15-mmol lꢀ1HP-
β-CD with 5% acetonitrile as the mobile phase. As
shown in Table S5, with the increase of flow rate, the
values of k, α, Rs, and N of these enantiomers gradually
decreased. It is well known that the theoretical plate
number is used to quantitatively represent the separa-
tion efficiency of chromatographic columns. Consider-
ing the impact of the above factors, this article chose
0.6 ml minꢀ1
.
3.1.4 | Influence of column temperature
Through the above experiments, the results showed
that the optical isomer of MA, MMA, and PMA can be
separated well under the selected chromatographic con-
ditions as shown in Figure 3.
In the above optimal mobile phase, the influence of
column temperature on enantioseparation of MA,

6
SHI ET AL.
described by the Gibbs–Holmholtz and van't Hoff equa-
tions as follows28,29
:
ΔH ΔS
RT
lnk ¼ ꢀ
þ
þlnϕ,
R
ð1Þ
ΔðΔHÞ ΔðΔSÞ
lnα ¼ ꢀ
þ
,
ð2Þ
ð3Þ
RT
R
ΔG ¼ ΔH ꢀTΔS,
where k is the capacity factor of enantiomers and α is the
separation factor between (R)-isomer and (S)-isomer. ΔG,
ΔH, and ΔS represent the Gibbs free energy change, the
enthalpy change, and the entropy change, respectively. φ
is the phase ratio. The experimental results showed that
there was a good linear correlation between lnk0 and 1/T
for these isomers (Figure 4), suggesting that the ΔH and
ΔS values were invariable over the entire temperature
range studied and the chiral recognition mechanism for
MA, MMA, and PMA under selected chromatographic
system was unchanged. To our knowledge, in the tradi-
tional reversed-phase chromatography system, (R/S)-
isomers have the same k values; thus, the Δ(ΔH) and
Δ(ΔS) values between (R/S)-isomers are zero. Neverthe-
less, as shown in Table 1, in the reversed-phase chro-
matographic system with HP-β-CD chiral additive, the
Δ(ΔH) and Δ(ΔS) values between (R/S)-isomers were not
zero, suggesting that the interaction of HP-β-CD with
(R)-isomers and (S)-isomers is different. Due to jΔ(ΔH)j
> jTΔ(ΔS)j,
it
can
be
deduced
that
the
enantioseparations of MA, MMA, and PMA are enthalpy-
driven processes. And we can also conclude that the chi-
ral recognition ability of HP-β-CD was in order of
FIGURE 3 Chromatograms of enantioseparations of
(A) mandelic acid, (B) 4-methoxymandelic acid, and
(C) 4-propoxymandelic acid enantiomers. Chromatographic
conditions: mobile phase: acetonitrile:PBS (0.1 mol lꢀ1, pH 3.00)
containing HP-β-CD (15 mmol lꢀ1) = 5:95 (v/v). Flow rate:
0.6 ml minꢀ1; detection wavelength: 230 nm; column temperature:
30ꢂC
3.2 | Assessment on thermodynamic
parameters
The relationship between the retention factor (k) or sepa-
ration factor (α) and the column temperature (T) during
chromatographic chiral separation process can be
FIGURE 4 van't Hoff plots of MA, MMA, and PMA

SHI ET AL.
7
TABLE 1 Regression parameters for Equation (1) and thermodynamic parameters
Regression parameters
for Equation (1)
ΔH
ΔS + Rlnφ
Δ(ΔH)b
Δ(ΔS)c
Δ(ΔG)d
Compounds
A
B
r2a
(kJ molꢀ1
)
(J molꢀ1 Kꢀ1
)
(kJ molꢀ1
)
(J molꢀ1 Kꢀ1
)
(J molꢀ1
)
(R/S)-MA
kR 1466.6 ꢀ3.8475 0.9986 ꢀ12.19
kS 1294.9 ꢀ3.4249 0.9987 ꢀ10.77
kR 1547.9 ꢀ3.8937 0.9977 ꢀ12.87
kS 1400.0 ꢀ3.5425 0.9980 ꢀ11.64
kR 1511.2 ꢀ1.6630 0.9974 ꢀ12.56
kS 1392.1 ꢀ1.4037 0.9977 ꢀ11.57
ꢀ31.99
ꢀ28.47
ꢀ32.37
ꢀ29.45
ꢀ13.83
ꢀ11.67
ꢀ1.42
ꢀ1.23
ꢀ0.99
ꢀ3.52
ꢀ2.92
ꢀ2.16-
ꢀ353.44
ꢀ345.24
ꢀ335.52
(R/S)-MMA
(R/S)-PMA
aRegression parameters for Equation (1), where A is equal to ΔH/R, B is equal to ΔS/R + lnφ, r is correlation coefficient, and R is gas constant (8.314 J molꢀ1).
bΔ(ΔH) is calculated by ΔHR and ΔHs.
cΔ(ΔS) is calculated by ΔSR + Rlnφ and ΔSs + Rlnφ.
dΔ(ΔG) is calculated by ΔG = ΔH ꢀ TΔS at 303 K.
MA > MMA > PMA. Additionally, the mode of interac-
tion between small molecule and macromolecule can be
deduced on the basis of the signs of the thermodynamic
parameters like ΔH and ΔS. Generally, the negative
values of both ΔH and ΔS represent that the main force
is van der Waals force and/or hydrogen-bonding interac-
tion. The positive values of both ΔH and ΔS mean that
the main interaction force is a hydrophobic interaction.
ΔH of almost zero and the positive value of ΔS represent
that the main interaction force is an electrostatic force.32
Therefore, we deduced that the main mode of the interac-
tion of HP-β-CD with MA, MMA, and PMA was van der
Waals and hydrogen-bonding interactions.
in Figure S1. As shown in Figure S1, the fluctuation
range of the RMSD value of each enantiomer is about
0.1 nm, which signifies the minimal deviation with stabi-
lized conformational coordinates.37 Subsequently, the
representative conformation with the lowest energy for
each complex was extracted by cluster analysis (cutoff
0.3 nm) during the 120-nsec dynamic simulations and
was represented in Figure 5. The findings revealed that
each enantiomer embedded into the hydrophobic cavity
of HP-β-CD. The phenyl, methoxyphenyl, and
propoxyphenyl groups were situated in the narrow edge
of HP-β-CD, whereas the –COOH moiety is situated in
the wider edge. And the insertion depth of (S)-isomer
was larger than that of (R)-isomer, suggesting that the
inclusion of (S)-isomer with HP-β-CD was stronger than
that of (R)-isomer.
3.3 | Molecular simulation
Additionally, the shape of HP-β-CD has been signifi-
cantly deformed as shown in Figure 5 and Table S6. And
the shape of the cavities in each inclusion complex is also
different. The conformational change can favor the inclu-
sion interaction between HP-β-CD and isomers to ensure
a better inclusion and enhance the complexation entropy.
This conformational change is defined as “induced-fit
mechanism.”38 In addition, the change was closely
related to the structure of guest molecules; that is, the
alteration in the conformation of HP-β-CD was different
for the different guest molecules. Thus, it can be inferred
that the flexibility of HP-β-CD played a dominative role
in the chiral recognition.
As mentioned in the assay of thermodynamic parameters,
the interaction between enantiomers and chiral selector,
HP-β-CD plays a significant role in the process of chiral
separation. The research on the chiral resolution mecha-
nism of enantiomers on CD and its derivatives has aroused
great attention.32–36 To clarify the chiral resolution mecha-
nism of (R/S)-MA, (R/S)-MMA, and (R/S)-PMA on HP-
β-CD, the inclusion of HP-β-CD with each optical isomer
was explored with help of molecular docking and MD
approaches. First, the inclusion between HP-β-CD and
each optical isomer was imitated using semiflexible dock-
ing, and thus, each enantiomer–HP-β-CD complex with
the lowest binding energy was obtained. To obtain the
more accurate geometry of each inclusion complex, each
enantiomer–HP-β-CD complex obtained from molecular
docking was used as initial structure for MD simulation.
In MD simulation, the root-mean-square deviation
(RMSD) for each enantiomer–HP-β-CD in complex was
calculated, and the results of MD simulation were shown
3.4 | Interaction forces between HP-
β-CD and isomers
As everyone knows, the inclusion interaction between
guest molecule and CDs includes various noncovalent

8
SHI ET AL.
FIGURE 5 Representative
geometries of complexes of HP-β-CD
with (A) mandelic acid,
(B) 4-methoxymandelic acid, and
(C) 4-propoxymandelic acid obtained by
MD simulation. (R)-1 and (S)-1 were
abstracted from the wider edge of the
HP-β-CD cavity, and (R)-2 and (S)-2
were abstracted from the side of the HP-
β-CD wall. R and S represented (R)-
isomer and (S)-isomer, respectively
TABLE 2 Various energies of various complexes obtained by molecular dynamics using MM-PBSA methoda
Complexes
ΔEvdw
ΔEelec
ΔEpolar
ΔEsasa
ΔEbinding
Δ(ΔEbinding)
(R)-MA–HP-β-CD
(S)-MA–HP-β-CD
ꢀ61.140 1.133 ꢀ15.017 0.439
ꢀ92.176 0.372 ꢀ15.637 0.316
ꢀ120.203 0.350 ꢀ10.616 0.301
42.442 0.698
54.537 0.349
ꢀ6.697 0.107 ꢀ40.312 1.043 22.111 0.73
ꢀ9.162 0.022 ꢀ62.423 0.304
(R)-4-MMA–HP-
β-CD
53.920 0.430 ꢀ10.517 0.027 ꢀ69.389 0.345
4.425 0.009
(S)-4-MMA–HP-
β-CD
ꢀ112.003 0.326 ꢀ18.942 0.371
67.972 0.428 ꢀ10.856 0.024 ꢀ73.814 0.354
(R)-4-PMA–HP-
β-CD
ꢀ113.647 1.141 ꢀ20.188 0.634 64.413. 0.634 ꢀ11.762 0.028 ꢀ81.169 0.391
ꢀ114.816 0.358 ꢀ16.631 0.303 60.240 0.341 ꢀ11.930 0.026 ꢀ83.151 0.344
1.982 0.047
(S)-4-PMA–HP-
β-CD
aΔEvdw denotes van der Waals energy; ΔEelec is electrostatic energy including all electron delocalizations; ΔEpolar is polar solvation energy; ΔEsasa is SASA
energy, where SASA is defined as a surface area of macromolecules that is accessible to aqueous solvent; ΔEbinding is binding energy; and Δ(ΔEbinding) is
calculated by ΔEbinding,R ꢀ ΔEbinding,S. All physical quantities are kJ molꢀ1 in units.
bond interactions. The interaction energies were
analyzed for each inclusion complex of 20- to 100-nsec
trajectory of MD simulation using MM-PBSA approach
in GROMACS and APBS1 software. Generally speaking,
the MM-PBSA method cannot be completely consistent
with the experimental results, but it can qualitatively
analyze the main sources of contribution in chiral recog-
nition.39 As is well known, in the process of the interac-
tions between host and guest, the main interaction forces
have van der Waals force, electron delocalization interac-
tion, solvent-accessible surface area (SASA) interaction,
and the polar solvation interaction. Electron
delocalization interactions include hydrogen-bonding
and dipole–dipole interaction. SASA energy represents
the transfer free energy required to transfer macromole-
cule from aqueous solvents to nonpolar solvents.37
From Table 2, it can be found that the interaction
forces included the contribution of various forces is dif-
ferent in each isomer and HP-β-CD interaction, the van
der Waals force, electron delocalization energy, and
SASA energy favored of the formation of inclusion com-
plex whereas the polar solvation energy was not favor to
inclusion interaction. This was consistent with the result
obtained by the analysis of thermodynamic parameters.

SHI ET AL.
9
The binding energy of HP-β-CD with (S)-isomer for three
chiral compounds is more negative than that with (R)-
isomer, suggesting that the interaction between HP-β-CD
and (S)-isomer was stronger than that of (R)-isomer,
resulting that the concentration of (S)-isomer in mobile
phase was larger than that of (R)-isomer, and thus, (S)-
isomer is first washed out. This is consistent with the
experimental results. Meanwhile, it can be also seen that
ORCID
Jie-Hua Shi https://orcid.org/0000-0002-9807-5654
REFERENCES
1. Maier NM, Franco P, Lindner W. Separation of enantiomers:
needs, challenges, perspectives. J Chromatogr A. 2001;906(1-2):
3-33.
2. Zhang J, Mao H, Li M, Su E. Cyclodextrin glucosyltransferase
immobilization on polydopamine-coated Fe3O4 nanoparticles
in the presence of polyethyleneimine for efficient
β-cyclodextrin production. Biochem Eng J. 2019;150:107264.
3. Saenger WR, Jacob J, Gessler K, et al. Structures of the com-
mon cyclodextrins and their larger analogues—beyond the
doughnut. Chem Rev. 1998;98(5):1787-1802.
4. Santana ACSGV, Nadvorny D, Passos TDD, Soares MFDLR,
Soares-Sobrinho JL. Influence of cyclodextrin on posaconazole
stability, release and activity: improve the utility of the drug.
J Drug Deliv Sci Tec. 2019;53:101153.
5. Shi X, Zhou Y, Liu F, Mao J, Zhang Y, Shan T. Modeling of chi-
ral gas chromatographic separation of alkyl and cycloalkyl
2-bromopropionates using cyclodextrin derivatives as station-
ary phases. J Chromatogr A. 2019;1596:161-174.
the Δ(ΔEbinding), (Δ(ΔEbinding) = ΔEbinding,R ꢀ ΔEbinding,S
)
value was in order of MA–HP-β-CD complex > MMA–
HP-β-CD complex > PMA–HP-β-CD complex, indicating
that the chiral recognition ability of HP-β-CD was in
order of MA > MMA > PMA. It was consistent with the
order of Δ(ΔG) values obtained from van't Hoff plot
(Table 1), suggesting that the MD may be used to predict
the chiral recognition ability of HP-β-CD and provide a
certain degree of reference for the identification of chiral
drugs in the future.
4 | CONCLUSION
6. Ye J, Yu W, Chen G, Shen Z, Zeng S. Enantiomeric separation
of 2-arylpropionic acid nonsteroidal anti-inflammatory drugs
by HPLC with hydroxypropyl-β-cyclodextrin as chiral mobile
phase additive. Biomed Chromatogr. 2010;24(8):799-807.
7. Huang Z, Guo D, Fan J, et al. HPLC semi-preparative separa-
tion of diclazuril enantiomers and racemization in solution.
J Sep Sci. 2020;43(7):1240-1247.
The enantioseparations and chiral recognition mecha-
nism of MA, MMA, and PMA were investigated by
reversed-phase HPLC using HP-β-CD as chiral selector
and by MD simulation approach. The results revealed
that there was a good separation efficiency between
(R/S)-isomers of the three compounds (Rs > 1.5) under
selected chromatographic conditions, indicating that HP-
β-CD possessed good chiral recognition ability for the
three chiral compounds. The chiral recognition was
responsible for forming inclusion complexes of (R/S)-
isomers with HP-β-CD with the different conformation
and binding energy. The binding energy of HP-β-CD with
(S)-isomer for three chiral compounds is more negative
than that with (R)-isomer, suggesting that the elution
order was (S)-isomer > (R)-isomer. It is proved that the
study on elution order of enantiomers through molecular
simulation approach is technically feasible. In the process
of the formation of inclusion complex, the main interac-
ꢀ
8. Casado N, Saz JM, García MA, Marina ML. Modeling-based
optimization of the simultaneous enantiomeric separation of
multicomponent mixtures of phenoxy acid herbicides using
dual cyclodextrin systems by capillary electrophoresis.
J Chromatogr A. 2019;1610:460552.
9. Deng M, Li S, Cai L, Guo X. Preparation of a hydroxypropyl-
β-cyclodextrin functionalized monolithic column by one-pot
sequential reaction and its application for capillary electro-
chromatographic enantiomer separation. J Chromatogr A.
1603;2019:269-277.
10. Tang K, Zhang P. Enantioselective extraction of terbutaline
enantiomers with β-cyclodextrin derivatives as hydrophilic
selectors. Chem Pap. 2011;65:273-279.
11. Hadjmohammadi MR, Hashemi M. Chiral separation of meth-
adone using solid membrane extraction based on chiral selec-
tor, solid membrane: sheep skin leather. J Iran Chem Soc. 2019;
16(8):1611-1616.
tion energies included ΔEvdw, ΔEelec, ΔEsasa, and ΔEpolar
.
Meantime, it can be revealed from the theoretical calcula-
tion results that the chiral recognition ability of HP-β-CD
was in order of MA > PMA > MMA. It was consistent
with the order of Δ(ΔG) values obtained from van't Hoff
plot (Table S6), suggesting that the MD may be used to
predict the chiral recognition ability of HP-β-CD.
12. Tang C-D, Shi H-L, Jia Y-Y, et al. High level and
enantioselective production of L-phenylglycine from racemic
mandelic acid by engineered Escherichia coli using response
surface methodology. Enzyme Microb Technol. 2020;136:
109513.
ꢀ
13. Blay G, Fernandez I, Monje B, Muñoz MC, Pedro JR, Vila C.
Enantioselective synthesis of 2-substituted-1,4-diketones from
(S)-mandelic acid enolate and α,β-enones. Tetrahedron. 2006;
62(39):9174-9182.
ACKNOWLEDGMENT
This study was funded by Huahai Pharmaceutical Co.,
Ltd. (YX-KF-[2005]018).
14. Zang J, Wang ZJ, Wu YZ. The optical resolution of mandelic
acid. Fine Chem Intermedia Tes. 2005;35:50-52.

10
SHI ET AL.
15. Hallas G, Yoon C. The synthesis and properties of
naphthodifuranones and naphthofuranonepyrrolidones. Dyes
Pigments. 2001;48(2):121-132.
16. Tong S, Zhang H, Shen M, Ito Y, Yan J. Enantioseparation of
mandelic acid derivatives by high performance liquid chroma-
tography with substituted β-cyclodextrin as chiral mobile phase
additive and evaluation of inclusion complex formation.
J Chromatogr B. 2014;962:44-51.
17. Tong SQ, Shen MM, Xiong Q, Wang XP, Lu MX, Yan JZ. Chiral
ligand exchange countercurrent chromatography: equilibrium
model study on enantioseparation of mandelic acid.
J Chromatogr A. 2016;1447:115-121.
29. Shi JH, Xu SX, Jia QQ, Yan XQ. Characterization of a cellulose
trisphenylcarbamate/1-octyl-3-methylimidazolium
tet-
rafluoroborate mixture as GC stationary phase: thermodynamic
parameters and LSER methodology. Chromatographia. 2013;
76(15-16):1021-1029.
30. Grillo R, de Melo NF, Moraes CM, et al. Study of the
interaction
between
hydroxymethylnitrofurazone
and
2-hydroxypropyl-β-cyclodextrin. J Pharm Biomed Anal. 2008;
47(2):295-302.
31. Rong R, Deng Y. Instrumental Analysis, Second Ed. Beijing:
China Medical Science Press; 2018. 180p.
32. de Miranda TM, de Oliveira AR, Pereira JR, et al. Inclusion
vs. micellization in the cethylpyridine chloride/β-cyclodextrin
system: a structural and thermodynamic approach. J Mo Struc.
2019;1184:289-297.
33. Lipkowitz KB. Applications of computational chemistry to the
study of cyclodextrins. Chem Rev. 1998;98(5):1829-1874.
34. Shi JH, Hu Y, Ding ZJ. Theoretical study on chiral recognition
mechanism of ethyl-3-hydroxybutyrate with permethylated
β-cyclodextrin. Comput Theor Chem. 2011;973(1-3):62-68.
35. Ali SM, Muzaffar S, Imtiaz S. Comparative study of complexa-
tion between cyclodextrins and xylazine using 1H NMR and
molecular modelling methods. J Mol Struct. 2019;1197:56-64.
36. Shi JH, Ding ZJ, Hu Y. Theoretical study on chiral recognition
mechanism of methyl mandelate enantiomers on
permethylated β-cyclodextrin. J Mol Model. 2012;18(2):803-813.
37. Kumari R, Kumar R, Open Source Drug Discovery Consortium,
Lynn A. g_mmpbsa—a GROMACS tool for high-throughput
18. Shi JH, Su YH, Jiang W. Enantioseparation and chiral recogni-
tion
of
α-cyclohexylmandelic
acid
and
methyl
α-cyclohexylmandelate on hydroxypropyl-β-cyclodextrin as chi-
ral selector: HPLC and molecular modelling. J Chromatogr Sci.
2013;51(1):8-16.
19. Shi JH, Ding ZJ, Hu Y. Experimental and theoretical studies on
the enantioseparation and chiral recognition of mandelate and
cyclohexylmandelate on permethylated β-cyclodextrin chiral
stationary phase. Chromatographia. 2011;74(3-4):319-325.
20. Zhang P, Feng X, Tang K, Xu W. Study on enantioseparation of
α-cyclopentyl-mandelic acid enantiomers using continuous
liquid-liquid extraction in centrifugal contactor separators:
experiments and modelling. Chem Eng Process. 2016;107:
168-176.
21. Kellici TF, Ntountaniotis D, Leonis G, et al. Investigation of
the interactions of silibinin with 2-hydroxypropyl-
β-cyclodextrin through biophysical techniques and computa-
tional methods. Mol Pharm. 2015;12(3):954-965.
MM-PBSA calculations.
J Chem Inf Model. 2014;54(7):
1951-1962.
22. Zhao Q, Gao H, Su Y, et al. Experimental characterization and
molecular dynamic simulation of ketoprofen-cyclodextrin com-
plexes. Chem Phys Lett. 2019;736:136802.
38. Saenger W, Noltemeyer M, Manor PC, Hingerty B, Klar B.
“Induced-fit”-type complex formation of the model enzyme
α-cyclodextrin. Bioorg Chem. 1976;5(2):187-195.
23. Tripathi MK, Yasir M, Gurjar VS, Bose P, Shrivastava ADR.
Insights from the molecular docking of hydrolytic products of
methyl isocyanate (MIC) to inhibition of human immune pro-
teins. Interdiscip Sci. 2015;7:287-294.
39. Meng X-Y, Zhang H-X, Mezei M, Meng C. Molecular docking:
a powerful approach for structure-based drug discovery. Curr
Comput Aided Drug des. 2011;7(2):146-157.
24. Wang JM, Wolf RM, Caldwell JW, Kollman PA, Case DA.
Development and testing of a general amber force field.
J Comput Chem. 2004;25(9):1157-1174.
SUPPORTING INFORMATION
Additional supporting information may be found in the
online version of the article at the publisher's website.
25. Andersen HC. Molecular dynamics simulations at constant
pressure and/or temperature.
2384-2390.
J Chem Phys. 1980;72(4):
How to cite this article: Shi J-H, Lin Z-Y,
Kou S-B, Wang B-L, Jiang S-L. Enantioseparation
of mandelic acid and substituted derivatives by
high-performance liquid chromatography with
hydroxypropyl-β-cyclodextrin as chiral mobile
additive and evaluation of inclusion complexes by
molecular dynamics. Chirality. 2021;1-10. doi:
10.1002/chir.23348
26. Wang R, Zhou H, Siu SWI, Gan Y, Wang Y, Ouyang D. Com-
parison of three molecular simulation approaches for
cyclodextrin-ibuprofen complexation. J Nanomate. 2015;2015:
193049.
27. Tang K, Chen Y, Huang K, Liu J. Enantioselective resolution of
chiral aromatic acids by biphasic recognition chiral extraction.
Tetrahedron. 2007;18(20):2399-2408.
28. Mcgachy NT, Grinberg N, Variankaval N. Thermodynamic
study of N-trifluoroacetyl-O-alkyl nipecotic acid ester enantio-
mers on diluted permethylated β-cyclodextrin stationary phase.
J Chromatogr A. 2005;1064(2):193-204.