Journal of Chromatography B, 1039 (2016) 79–87

Contents lists available at ScienceDirect

Journal of Chromatography B
journal homepage: www.elsevier.com/locate/chromb

Saponins from Panax bipinnatifidus Seem.: New strategy of extraction,

isolation, and evaluation of tyrosinase inhibitory activity based on
mathematical calculations
Yuchi Zhang a , Chong Shi b , Chunming Liu a,∗ , Min Yu c,∗ , Yanjuan Qi a , Sainan Li a
a
Central Laboratory, Changchun Normal University, No. 677 North Changji Road, Erdao District, Changchun 130032, China
b
Surgical Department, The Affiliated Hospital to Changchun University of Chinese Medicine, No. 1478 Gongnong Road, Chaoyang District, Changchun
130021, China
c
Department of Nephrology, Affiliated Hospital, Academy of Military Medical Sciences, No. 8 East Avenue, Fengtai District, Beijing 100071, China

a r t i c l e i n f o a b s t r a c t

Article history: A hyphenated accelerated solvent extraction (ASE) technique coupled with centrifugal partition chro-
Received 23 September 2016 matography (CPC), ultra high performance liquid chromatography (UPLC), and mass spectrometry (MS)
Received in revised form 21 October 2016 was established. The CPC fractions were prepared using the hyphenated technique. Subsequently, tyrosi-
Accepted 30 October 2016
nase inhibitory activities of the CPC fractions were experimentally evaluated, and the activities of
Available online 1 November 2016
individual components were calculated theoretically. This new approach was applied to saponin fractions
obtained from 10.0 g of raw Panax bipinnatifidus Seem. via a biphasic solvent system of ethyl acetate/n-
Keywords:
butanol/methanol/water (2:3:2:5, v/v/v/v). The CPC fractions monitoring was performed using an online
Accelerated solvent extraction
Centrifugal partition chromatography
UPLC/PDA system at 5-min intervals. The tyrosinase inhibitory activities of all fractions were analysed
Tyrosinase inhibitor using the fluorescence method. Mathematical calculations indicated that the inhibition rates of the gin-
Mathematics senosides Rh1 , Rh2 , Rg1 , Rg2 , and chikusetsusaponin L5 were all above 50.00%, showing potential for
Panax bipinnatifidus further development. The results were confirmed by comparison with authentic standards.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction pure compound. Separation of all compounds in medicinal plant

Medicinal herbs are the main resources for drug and health
food development [1,2]. Therefore, the extraction, fractionation,
and screening of their bioactive chemical constituents are challeng-
ing [3]. Since a large number of individual compounds is usually
present in medicinal herbs, with only some of them being bioac-
tive [4], it is necessary to establish a method for fast differentiation
between active and inactive compounds. The conventional active
chemical components screening procedure is as follows: the raw
materials are extracted, the extracted solution is concentrated and
isolated, each fraction is analysed, and the fractions containing
the pure compound are enriched for activity evaluation [5,6]. The
above process is mature and effective, with a very large application
range. However, it also has some limitations. For example, chro-
matographic separation of natural compounds makes it is difficult
to obtain fractions or compounds with a purity of more than 98%
[7,8], complicating the accurate evaluation of the activity of each
∗ Corresponding authors.
E-mail addresses: chem lab@sina.cn (C. Liu), papaym@163.com (M. Yu).
extracts is a waste of manpower, material resources, and time.
Therefore, the isolation and activity evaluation of all compounds
in medicinal herbs is difficult. To solve this problem, mathematical
methods were employed for the characterisation of the chromato-
graphic fractions and their activities. Fits of the functional relations
of the chemical compositions and fraction activities were obtained,
and the activity of each compound in the fraction was calculated.
In order to quickly obtain chromatographic fractions and anal-
yse their compositions, a fast integrated technique of extraction,
fractionation, and analysis was necessary. Conventional thermal
extraction methods such as reflux and decoction may cause a loss of
target compounds due to their ionisation, hydrolysis, and oxidation
during extraction [9]. The filtration and concentration of the extract
are labour-intensive and time-consuming. To obtain more effec-
tive constituents from raw medicinal herbs, accelerated solvent
extraction (ASE) could be used [10]. Subsequently, counter-current
chromatography (CCC) and centrifugal partition chromatography
(CPC) support-free liquid-liquid partition chromatographic tech-
niques can be employed to separate the target compounds or
certain fractions from the extract on a large scale, eliminating irre-
versible adsorption of the sample onto the solid support [11,12].

http://dx.doi.org/10.1016/j.jchromb.2016.10.043
1570-0232/© 2016 Elsevier B.V. All rights reserved.
80 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87
CPC is based on hydrostatics, helping avoid emulsification when
using n-butanol and/or isopropanol and hence increasing the reten-
tion rate of the stationary phase. This technique has been widely
used in the preparative separation and purification of saponins [13].
According to previous research and/or general experiments, the
fractions collected during preparative isolation should be analysed
by high performance liquid chromatography (HPLC) and/or mass
spectrometry (MS) to evaluate their purity and structural informa-
tion. In some cases, it may be necessary to dry-concentrate the CPC
fractions and re-dissolve their residues in appropriate HPLC sol-
vents prior to analysis. These additional steps are time-consuming
and increase the risk of target compound degradation (e.g., oxida-
tion). To obtain information on fraction composition avoiding these
post-isolation steps, an aliquot of the CPC effluent is injected into
an online HPLC or MS system for instantaneous analysis [14].
In addition, the extraction and separation processes are
performed independently. Following the extraction, it may be nec-
essary to dry the extracted solution and re-dissolve the residue in
appropriate solvents prior to separation. To expedite these non-
automatic and non-industrial production processes, integrated
techniques [15] together with automated extraction and separation
processes [16,17] have been used for the extraction and isolation
of natural products. Herein, a novel automated ASE method cou-
pled with online CPC and UPLC/PDA was established for continuous
extraction, online isolation, purity analysis, and structural identi-
fication of chemical constituents from raw plant materials (Fig. 1).
The CPC stationary phase was also used as the ASE extraction sol-
vent. Following extraction, the extracted solution was pumped into
the sample loop and then into the CPC column as a stationary phase,
and the target compounds were eluted with the mobile phase. CPC
fractions were collected using the six-port valve of the sample loop,
and their purity was analysed by online UPLC/PDA.
Panax bipinnatifidus Seem., belonging to the Panax family, is
a medicinal herb widely used in Vietnam for improving cogni-
tive processes and memory, reducing cancer risk, and lowering
blood sugar levels in diabetics [18,19]. Saponins including gin-
senosides, notoginsenosides, and chikusetsusaponins are the major
active compounds in all parts of the herb. The ginsenosides are
one of the most important compound types in the Panax family,
widely used for cancer treatment [20], improving skin functions
[21], combating aging and central nervous system disorders [22],
improving cardiovascular activity [23], and etc. The notoginseno-
sides and chikusetsusaponins are other saponin types found in
P. bipinnatifidus Seem., acting as immunological adjuvants [24,25]
and anti-obesity agents [26,27]. The main chemical constituent of
this medicinal herb is similar to that of ginseng and notoginseng.
However, it has a relative low price and can be used as a food supple-
ment in the southwest of China and Vietnam, showing considerable
development potential.
The activity screening of medicinal plant extracts showed
that the extracts of P. bipinnatifidus Seem. exhibit strong tyrosi-
nase inhibitory activity. Therefore, the abovementioned herb was
investigated to determine the tyrosinase inhibitory activities of
monomeric compounds. However, P. bipinnatifidus Seem. con-
tains more than 20 compounds, which are difficult to separate.
In nowadays, “Receptor-ligand” affinity experiments based on the
ultrafiltration liquid chromatography combine with mass spec-
trometry have a wide range applications [28,29]. However, the
method mentioned above also has flaws. For example, the chemical
compounds and ultrafiltration membrane has non-specific bind-
ing and then produce false positive results, hardly to provided
the inhibition rate of each compounds [30,31]. Therefore, the cur-
rent activity detection kits are still mainly based on the spectral
method [32–34]. Thus, mathematical analysis was employed to
evaluate the activities of non-isolated compounds, since calculating
the activities of all compounds in the extract does not require their
separation. This idea can provide a new strategy and methodology
for the analysis of compound activities in complex medicinal herb
systems. In addition, the common saponin detection wavelengths
are 200–203 nm, with the CPC solvent system possibly affecting
UV absorption and, therefore, the analysis results. In view of the
above, an evaporative light scattering detector (ELSD) is usually
used for the detection of saponins [35]. However, the CPC eluents
cannot be enriched for bioactivity screening, since the chemicals
are destroyed by ELSD. Using online UPLC/PDA to detect the CPC
eluent has the advantage of low sample consumption, not requiring
ELSD and being very suitable for this study.
The aims and achievements of this work are as follows:
1. Establish a new hyphenated ASE strategy with CPC and
UPLC/PDA (ASE/CPC/UPLC/PDA) coupled online to the extraction,
online separation, and analysis of the fractions from P. bipinnat-
ifidus Seem.
2. Calculate the compound activities by analysing the composition
and activities of CPC fractions, not requiring the separation of all
compounds.
The method established in this study has advantages in the eval-
uation of activities of complex compounds, being suitable for the
analysis and screening of active components from medicinal herbs.
2. Experimental
2.1. Reagents and raw materials
n-Butanol, ethyl acetate, and methanol used were of analyti-
cal grade (Beijing Chemicals, Beijing, China). Water was purified
using a Milli-Q water purification system (Millipore, Boston, USA).
Acetonitrile was of HPLC grade (Fisher Scientific, Shanghai, China).
Panax bipinnatifidus was purchased from the Hebei YuYanTang
Medicinal Store and identified by Dr. Yuchi Zhang (Changchun
Normal University, Changchun, China). tyrosinase assay kit was
purchased from Sigma (St. Louis, MO, USA).
2.2. Apparatus
An accelerated solvent extraction 150 system (Dionex, Sunny-
vale, CA, USA) with a 100-mL stainless steel ASE vessel was used.
Centrifugal partition chromatography was performed on an SIC
CPC-240 high performance centrifugal partition chromatography
system (System Instruments Co., Ltd., Japan) that was modified
in our laboratory. The CPC column is a stacked circular partition
disk rotor containing 2136 partition cells with a total internal vol-
ume of about 240 mL. The column is connected to the injector and
the detector via two high-pressure rotary seals. A four-port valve,
integrated into the CPC apparatus, allows operation of the col-
umn in either ascending or descending mode, depending on the
relative density of the mobile and stationary phases. Ultra high-
performance liquid chromatography (UPLC) was carried out on a
Waters Acquity H class instrument coupled with a Waters diode
array detector (Milford, USA).
2.3. Online ASE/CPC/UPLC/PDA procedure
Samples were ground in a high-speed disintegrator (Model
SF-2000, Chinese Traditional Medicine Machine Works, Shang-
hai, China) to obtain a fine powder (particle size of 0.6 mm). The
dry powder (10.0 g) was put into the 100-mL ASE vessel, and the
reagents were pumped into the mixer & degasser at a certain flow
rate. The mixed solvent was separated into two layers in an ASE sol-
vent bottle, and the four-port valve was turned to interlink ports 1/2
 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87
Fig. 1. Diagram of the instrumental setup enabling hyphenation of ASE, CPC, and UPLC/PDA.
and 3/4, while the left six-port valve was turned to interlink ports
1/6, 2/3, and 3/4. The lower phase in the solvent bottle was then
used to extract the raw materials (extraction time = 5 min, num-
ber of extractions = 1, extraction temperature = 60 ◦ C). During the
extraction, the upper phase in the ASE solvent bottle was pumped
into the CPC column using pump 1, and the CPC column was rotated
at 1100 rpm for 10 min. Subsequently, the four-port valve was
turned to interlink ports 1/4 and 2/3, while the lower phase in the
solvent bottle was pumped into the CPC column at a flow rate of
2.0 mL/min. After ASE, the extraction solution was flowed into the
sample loop via the left six-port valve (ports 1/6, 2/3, and 3/4 were
still linked), which was then turned to link ports 1/2, 3/4, and 5/6,
and the extraction solution in the sample loop was pumped into the
CPC column via pump 1. The CPC eluent subsequently flowed out
and was monitored by the UPLC/PDA system. During the UPLC/PDA
operation, the CPC separation was still in progress, and the frac-
tions could still fill the injection loop. After the UPLC/PDA analysis
was completed, the other fractions in the injection loop were ana-
lysed, and the operation was repeated. The UPLC chromatograms
were recorded and analysed, while the CPC fractions were collected,
concentrated to dryness by a termovap sample concentrator, and
stored at −4 ◦ C for activity evaluation.
2.4. UPLC/PDA analysis
CPC fractions were collected in 5-min intervals. Each fraction
was analysed by online UPLC. The analysis was performed on a
Waters SunFireTM C18 column (4.5 ␮m, 250 mm × 2.1 mm) at 30 ◦ C.
Acetonitrile (phase A) and water were used as the mobile phase in
a gradient elution mode of 0–4 min, 25%–70% (phase A). The mobile
phase flow rate was 0.4 mL/min. The effluents were monitored
at 203 nm using a photodiode array detector. MS/MS detection
was performed in the positive ionisation mode, with the following
parameters: source temperature = 120 ◦ C, desolvation tempera-
ture = 350 ◦ C, capillary voltage = 1.0 kV, and desolvation and cone
gas flows = 700 and 150 L/h, respectively. Nitrogen was used as a
collision gas, with a flow rate of 0.15 mL/min.
2.5. Selection of the CPC solvent system
The composition of the two-phase solvent system was selected
according to the partition coefficient (K) of crude sample tar-
get compounds. Approximately 3 mg of the crude sample were
weighed in a test tube, to which 2 mL of each phase of the
equilibrated biphasic solvent system were added. The tube was
81
vigorously shaken to equilibrate the sample thoroughly between
the two phases, and each phase was separated and evaporated to
dryness under N2 gas. The residue was dissolved in methanol and
analysed by UPLC. The partition coefficient (K) was expressed as
the peak area of the target compound in the upper phase divided
by that in the lower phase [36]. In this study, CPC fractions were
analysed instead of single compounds, therefore, the solvent sys-
tem capable of eluting all compounds from the crude extract was
chosen.
2.6. Tyrosinase inhibition assay
The CPC fractions were tested for tyrosinase inhibition abil-
ity using a previously described spectrophotometric method with
slight modifications [37]. Briefly, 140 ␮L of the assay mixture con-
taining 20 ␮L of sample solution, 20 ␮L of 250 U/mL tyrosinase
solution, and 100 ␮L of 100 mM phosphate buffer were added to
each well of a 96-well plate. The microplate was pre-incubated at
25 ◦ C for 10 min. Subsequently, 20 ␮L of 3 mM l-tyrosine solution
were added. After 30 min of incubation at 25 ◦ C, the absorbance
of the mixture was determined at 492 nm using an TECAN A-5082
microplate reader instrument (Tecan Group Ltd., Austria). Arbutin
was used as the positive control, and the assay was repeated four
times. The percentage inhibition of tyrosinase activity was calcu-
lated by Eq. (1):
Inhibition% = [1 − (A1 − A2 )/(A3 − A4 )] × 100% (1)
where A1 is the absorbance of the test samples and enzyme at
492 nm, A2 is the absorbance of the test samples without the
enzyme at 492 nm, A3 is the absorbance of the enzyme without
the test sample at 492 nm, and A4 is the absorbance of the solution
without the test samples and enzyme at 492 nm.
2.7. Calculation and optimisation of CPC composition and
tyrosinase inhibitio
The inhibition of tyrosinase depends on the compound con-
centrations in CPC fractions. Since CPC fractions contain 3–4
compounds, their concentrations were variable. Therefore, the
corresponding tyrosinase inhibition abilities were taken as an inde-
pendent variable, with the inhibition abilities of compounds 1, 2,
3, and 4 in the CPC fractions defined as x1 , x2 , x3 , and x4 , respec-
tively. The tyrosinase inhibition data were defined as the dependent
variable.
82 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87
Therefore, different multivariate function models were used to
analyse the relationship between independent and dependent vari-
ables and to subsequently determine the former. For example, the
tyrosinase inhibition ability of compound 1 was denoted as x1 , that
of compound 2 was denoted as x2 , and those of compounds 3 and 4
were denoted as x3 and x4 , respectively, etc. The tyrosinase inhibi-
tion data of CPC fraction 1 were denoted as y1 , those of CPC fraction
2 as y2 , etc. With the help of linear and/or nonlinear regression anal-
ysis, the dependences of y1 , y2 , y3 ,. . ., yn on x1 , x2 , x3 ,. . ., xn can be
well expressed as (logc1,1 ) · x1 + (logc1,2 ) · x2 + (logc1,3 ) · x3 + (logc1,n )
· xn = y1 , proven by testing the least-square error. The functions for
CPC fractions (1–m) were expressed as follows:
(logc 1,1 )·x1 + (logc 1,2 )·x2 + (logc 1,3 )·x3 + (logc 1,n )·xn = y1
(logc 2,1 )·x1 + (logc 2,2 )·x2 + (logc 2,3 )·x3 + (logc 2,n )·xn = y2
(logc 3,1 )·x1 + (logc 3,2 )·x2 + (logc 3,3 )·x3 + (logc 3,n )·xn = y3
..
.
(logc m,1 )·x1 + (logc m,2 )·x2 + (logc m,3 )·x3 + (logc m,n )·xn = ym
The log c1,1 , c1,2 , c1,n , c2,1 , c2,2 , c2,n ,. . ., cm,1 , cm,2 , cm,n , y1 , y2 ,
y3 ,. . ., ym were determined experimentally, allowing the values of
x1 , x2 ,. . ., xn to be calculated.
3. Results and discussion
3.1. UPLC–MS analysis of the ASE extract of P. bipinnatifidus
Seem.
UPLC and MS data provided useful information for the compo-
nent identification in P. bipinnatifidus Seem. A total of 21 UPLC peaks
were identified by comparing their retention time (tR ), molecule
peaks ([M+Na]+ ), and MS/MS fragment ions with those of authentic
standards reported in literature [38,39] (Table 1). Further UPLC/MS
and MS/MS data for the identification of the structures of the 21
compounds are shown in Table 1.
The MS data for peaks 1 and 6 are shown in Table 1. The
respective parent [M+Na]+ ions at m/z 809 and 823 correspond
to compound molecular weights of 786 and 800, respectively. The
MS/MS data of the ion at m/z 809 is shown Table 1, with all fragment
ions in the MS/MS spectrum produced directly from the parent ion.
Three main fragment ions at m/z 791, 749, and 677 were observed.
The mass differences between the parent ion and the fragment ions
at m/z 791 and 749 were 18 and 60, respectively, corresponding
to the neutral loss of water and a C3 H8 O unit. The mass differ-
ences between the parent ion and the fragment ion at m/z 677 was
132, corresponding to the neutral loss of an arabinose unit. One
main fragment ion at m/z 643 was observed. The mass difference
between the parent ion and the fragment ion was equal to 180, cor-
responding to the neutral loss of water and a hexose unit. Based on
comparison of MS/MS data and UPLC retention time with those of
the authentic standard, the compounds represented by peaks 1 and
6 were identified as majonoside R2 and majoroside F4 , respectively
[38].
The [M+Na]+ molecular ions of peaks 2, 3, 11, and 14 (Table 1)
were at m/z 985, 955, 793, and 1263, respectively, corresponding
to molecular weights of 962, 932, 770, and 1241, respectively. The
main fragment ions of m/z 985, 955, 793, and 1263 had m/z 805, 775,
661, and 1131, respectively, corresponding to neutral losses of 180,
180, 132, and 132 Da (one unit of glucose and xylose). The observed
and calculated mass errors were calculated as 2.03, −1.26, 3.27, and
−4.99, respectively, indicating that the calculation was accurate.
The compounds represented by peaks 2, 3, 11, and 14 were there-
fore identified as notoginsenoside R6, R1 , R2 , and Fa , respectively, by
comparing their retention time and mass spectral data with those
of the authentic standard, as well as on the basis of observed and
calculated mass [38].
The [M+Na]+ molecular ions of peaks 7, 8, 12, and 20 (Table 1)
were at m/z 677, 939, 925, and 939, respectively, corresponding
to molecular weights of 654, 916, 902, and 916, respectively. The
main fragment ions for m/z 677, 939, 925, and 939 were m/z 497,
807, 745, and 807, respectively, corresponding to neutral losses of
180, 132, 180, and 132 Da (glucosyl or xylosyl unit). The compounds
represented by peaks 7, 8, 12, and 20 were therefore identified as
vina-ginsenoside R10 , chikusetsusaponin III, chikusetsusaponin L5 ,
and chihusetsusaponin [38,39], respectively, by comparing their
retention time and mass spectral data with those of the authentic
(2)
standard.
(3) Using the same method, we assigned the compounds repre-
sented by peaks 4, 5, 9, 10, 13, 15–19, and 21 to ginsenoside Re,
(4)
Rg1, Rf, Ra3, Rb1, Rh2, Rg2, Rb2, Rb3, Rd, and Rh1, respectively, by
comparing the reference data with those of authentic standards.
3.2. Selection of CPC solvent and the procedure of extraction,
isolation, and collection
(5)
The CPC solvent system was also used as the ASE solvent. The
solvent system used should meet the requirements of both CPC
and ASE. General solvent systems for the separation of saponins are
chloroform/methanol/water, dichloromethane/methanol/water,
n-hexane/n-butanol/ethanol/water, ethyl acetate/n-
butanol/ethanol/water, etc. Therefore, the solvent systems
mentioned above were tested to determine ASE rates. The
amounts of target saponins obtained with different extrac-
tion solvents are listed in Table 2. Good reproducibility was
obtained irrespective of the ASE solvent used, and the cal-
culated standard deviation (S.D.) did not exceed 7.0%. The
solvent system was selected based on the obtained yield
(amount of saponins obtained from 100 g of crude plant mate-
rial). If the organic phases of chloroform/methanol/water and
dichloromethane/methanol/water at different volume ratios were
used as extraction solvents, the amounts of extracted saponins
were relatively low (28.54–37.12 mg/100 g, data not shown). If
the organic phases of n-hexane/n-butanol/ethanol/water and/or
ethyl acetate/n-butanol/ethanol/water at different volume ratios
were used as extraction solvents, the saponin yields dramatically
increased, reaching 68.54–156.21 mg/100 g. Since the toxicities
of chloroform and dichloromethane are relatively large, and
the extraction abilities of the solvent systems containing them
were low, n-hexane/n-butanol/ethanol/water and ethyl acetate/n-
butanol/ethanol/water systems were chosen as the extraction
solvent and were used for evaluating the partition coefficient (K).
The K values of saponins from P. bipinnatifidus Seem.
for n-hexane/n-butanol/ethanol/water and ethyl acetate/n-
butanol/ethanol/water systems are shown in Table 2.
When n-hexane/n-butanol/ethanol/water (volume ratios of
1:2:1:2, 2:2:2:2, 3:2:3:2, and 4:2:4:2) and ethyl acetate/n-
butanol/ethanol/water (volume ratios of 2:2:2:3, 3:2:2:4, 4:2:2:5,
5:2:2:6, and 6:2:2:7) were used as the solvent system, ginseno-
side Rd, ginsenoside Rh1 , and other saponins with polarities
between those of the above compounds exhibited relatively
large K values, dramatically prolonging the separation time and
making low-polarity saponins hardly possible to elute. When
n-hexane/n-butanol/ethanol/water (volume ratios of 6:2:2:6 and
7:2:2:7) and ethyl acetate/n-butanol/ethanol/water (volume ratios
of 9:2:2:9 and 10:2:2:10) were used as the solvent system, the
K values of majonoside R2 , ginsenoside Re, vina-ginsenoside R10 ,
and other saponins with polarities between those of the above
compounds were too low, resulting in a short total separation time,
reduced number of collected fractions (fractions were collected
 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87 83

Table 1
Characterisation of saponins from the ASE extract of Panax bipinnatifidus Seem.

NO. tR (min) Molecular weight Formula [M+Na]+ m/z UPLC/ESI/MS/MS data (m/z) Compounds identified

1 0.50 786 C41 H70 O14 809 MS/MS: 791 (100%), 749 (42%), 677 (31%) majonoside R2
2 0.56 962 C48 H82 O19 985 MS/MS: 365 (100%), 805 (62%), 305 (15%) notoginsenoside R6
3 0.65 932 C47 H80 O18 955 MS/MS: 775 (100%) notoginsenoside R1
4 0.69 946 C48 H82 O18 969 MS/MS: 789 (100%), 643 (46%), 441 (12%) ginsenoside Re
5 0.70 800 C42 H72 O14 823 MS/MS: 643 (100%) ginsenoside Rg1
6 1.21 800 C42 H72 O14 823 MS/MS: 643 (100%) majoroside F4
7 1.44 654 C36 H62 O10 677 MS/MS: 497 (100%), 659 (12%), 617 (35%), 586 (22%) vina-ginsenoside R10
8 1.76 916 C47 H80 O17 939 MS/MS: 497 (100%), 807 (56%) chikusetsusaponin III
9 1.96 800 C42 H72 O14 823 MS/MS: 641 (100%), 365 (51%), 661 (43%) ginsenoside Rf
10 2.04 1240 C59 H100 O27 1263 MS/MS: 497 (100%), 437 (42%), 789 (25%) ginsenoside Ra3
11 2.16 770 C41 H70 O13 793 MS/MS: 335 (100%), 661 (45%) notoginsenoside R2
12 2.21 902 C46 H78 O17 925 MS/MS: 467 (100%), 745 (21%) chikusetsusaponin L5
13 2.26 1108 C54 H92 O23 1131 MS/MS: 789 (100%), 425 (19%), 365 (8%) ginsenoside Rb1
14 2.36 1240 C59 H100 O27 1263 MS/MS: 1131 (100%), 921 (41%), 1090 (24%) notoginsenoside Fa
15 2.41 622 C36 H62 O8 645 MS/MS: 614 (110%), 585 (41%), 562 (35%), 545 (16%) ginsenoside Rh2
16 2.50 784 C42 H72 O13 807 MS/MS: 661 (100%), 481 (47%), 349 (11%) ginsenoside Rg2
17 2.54 1078 C53 H90 O22 1101 MS/MS: 789 (100%), 335 (18%) ginsenoside Rb2
18 2.73 1078 C53 H90 O22 1101 MS/MS: 789 (100%), 1079 (12%) ginsenoside Rb3
19 2.89 946 C48 H82 O18 969 MS/MS: 789 (100%), 425 (41%), 365 (30%), 587 (12%) ginsenoside Rd
20 3.32 916 C47 H80 O17 939 MS/MS: 807 (100%), 497 (18%) chihusetsusaponin
21 3.59 638 C36 H62 O9 661 MS/MS: 578 (100%), 481 (34%), 413 (12%) ginsenoside Rh1

Table 2
Description of ASE extracts and partition coefficient values for the solvent systems used.

No. Solvent system (v:v:v:v) Extraction rate a (mg/100 g) Extraction rate b (mg/100 g) Partition coefficient values of the interval compounds

Com.1c Com.2 Com.3 Com.4 Com.5 Com.6 Com.7 Com.8

1 n-hexane–n-butanol–ethanol–water (1:2:1:2) 132.75 ± 7.65 68.54 ± 4.23 1.45 2.31 1.95 3.65 7.25 5.65 12.65 19.32
2 n-hexane–n-butanol–ethanol–water (2:2:2:2) 125.9/ ± 6.24 75.25 ± 5.36 0.77 1.47 1.29 2.32 4.25 3.65 8.54 12.21
3 n-hexane–n-butanol–ethanol–water (3:2:2:3) 112.78 ± 6.48 80.69 ± 6.65 0.42 1.12 0.82 1.41 3.21 2.54 5.26 7.62
4 n-hexane–n-butanol–ethanol–water (4:2:2:4) 105.68 ± 6.65 85.69 ± 5.63 0.24 0.52 0.39 1.02 2.54 1.54 3.85 4.21
5 n-hexane–n-butanol–ethanol–water (5:2:2:5) 97.45 ± 5.32 92.54 ± 6.41 0.11 0.32 0.16 0.74 1.42 1.02 2.05 2.63
6 n-hexane–n-butanol–ethanol–water (6:2:2:6) 90.35 ± 6.35 105.42 ± 6.32 0.04 0.21 0.08 0.42 0.95 0.74 1.21 1.71
7 n-hexane–n-butanol–ethanol–water (7:2:2:7) 85.68 ± 5.95 101.21 ± 6.95 0.02 0.11 0.03 0.22 0.42 0.30 0.64 1.12
8 ethyl acetate–n-butanol–ethanol–water (2:2:2:3) 156.21 ± 6.56 74.53 ± 5.65 2.04 4.21 3.35 6.32 9.24 8.54 16.51 19.40
9 ethyl acetate–n-butanol–ethanol–water (3:2:2:4) 135.45 ± 6.45 78.69 ± 5.54 1.52 2.95 2.33 4.13 6.92 6.33 11.44 14.62
10 ethyl acetate–n-butanol–ethanol–water (4:2:2:5) 124.24 ± 6.65 85.36 ± 6.12 1.03 2.26 1.75 3.03 5.32 4.32 8.65 10.51
11 ethyl acetate–n-butanol–ethanol–water (5:2:2:6) 112.57 ± 6.59 90.25 ± 5.74 0.76 1.66 1.25 2.34 3.65 2.65 5.77 8.32
12 ethyl acetate–n-butanol–ethanol–water (6:2:2:7) 108.24 ± 5.41 97.65 ± 5.84 0.52 1.23 0.95 1.73 2.85 1.77 3.62 6.01
13 ethyl acetate–n-butanol–ethanol–water (8:2:2:8) 98.65 ± 5.56 106.54 ± 6.57 0.29 0.66 0.54 0.86 1.42 1.02 1.67 3.12
14 ethyl acetate–n-butanol–ethanol–water (9:2:2:9) 93.85 ± 5.35 101.42 ± 6.14 0.21 0.54 0.43 0.75 0.99 0.51 1.02 2.06
15 ethyl acetate–n-butanol–ethanol–water (10:2:2:10) 90.32 ± 6.21 98.13 ± 5.96 0.15 0.45 0.34 0.62 0.71 0.41 0.85 1.46
a
Upper phase of the solvent systems were used as the extraction solvent, data are expressed as mean ± SD, For each sample n = 3.
b
Lower phase of the solvent systems were used as the extraction solvent, data are expressed as mean ± SD, For each sample n = 3.
c
Com. 1: majonoside R2 ; Com. 2: ginsenoside Re; Com. 3: vina-ginsenoside R10 ; Com. 4: ginsenoside Ra3 ; Com. 5: ginsenoside Rb1 ; Com. 6: ginsenoside Rg2 ; Com. 7:
ginsenoside Rd; Com. 8: ginsenoside Rh1 .

in 5-min intervals), and increased analytical error. In view of
the above, n-hexane/n-butanol/ethanol/water (volume ratio of
4:2:2:4) and ethyl acetate/n-butanol/ethanol/water (volume ratio
of 8:2:2:8) systems met the requirements. The K values of the
target compounds were between 0.29 and 3.12 for ethyl acetate/n-
butanol/ethanol/water (volume ratio of 8:2:2:8), much better
than the K values of 0.24 to 4.21 for the other solvent system. To
collect enough fractions, decrease the analysis error, and shorten
the separation time, the ethyl acetate/n-butanol/ethanol/water
system (volume ratio of 8:2:2:8) was selected.
3.3. CPC fraction analyses, activity evaluation, and mathematical
calculations
After the extraction, the ASE extract was pumped into the CPC
column for fractionation, during which it was continuously moni-
tored by online UPLC/PDA until six consecutive CPC fractions were
obtained where no compounds could be detected. A total of 103
valid fractions were collected in this study.
All fractions were analysed by online UPLC/PDA; according to
the saponin composition, the fractions were divided into eleven
groups. The results shown in Fig. 2 and Table S1 (Supplementary
information) indicate that only two groups contained pure com-
pounds: group 6 contained ginsenoside Rb1, group 10 contained
ginsenoside Rd, with the other ones being mixtures. In addition,
this result also indicates that different groups contained the same
ingredients. For example, groups 1, 2, and 3 contained ginseno-
side Rg1; groups 1 and 2 contained ginsenoside Re and majoroside
Rg1 ; groups 2 and 3 contained majoroside F4 ; groups 3, 4, 5, and 6
contained ginsenoside Rf; groups 5, 6, 7, and 8 contained ginseno-
side Ra3 , etc. The main reason of the above mixing is the collection
of the whole CPC eluent, with 21 compounds being eluted in one
operation. In addition, we collected the whole CPC eluent in order to
minimise calculation errors, and to be able to mutually cross-verify
the calculation results of different groups with the same compound.
The group 1 contained fractions 1–7, with the chemical com-
pounds being majonoside R2 , notoginsenosides R6 and R1 , and
ginsenosides Re and Rg1 . The UPLC peak areas of the abovemen-
tioned saponins in fractions 1–7 are shown in Table 3.
The analysis of inhibition activities of the fractions indicates
that their concentrations were relative large, fractions 1–5 being
brownish black, and fractions 6 and 7 being dark brown. The inhi-
84 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87

Fig. 2. UPLC/PDA/MS chromatogram of Panax bipinnatifidus Seem. and group assignment according to the compounds present in the CPC eluent, with colour/grey used to
distinguish compound polarity.
1: majonoside R2 ; 2: notoginsenoside R6 ; 3: notoginsenoside R1 ; 4: ginsenoside Re; 5: ginsenoside Rg1 ; 6: majoroside F4 ; 7: vina-ginsenoside R10 ; 8: chikusetsusaponin III;
9: ginsenoside Rf; 10: ginsenoside Ra3 ; 11: notoginsenoside R2 ; 12: chikusetsusaponin L5 ; 13: ginsenoside Rb1 ; 14: notoginsenoside Fa; 15: ginsenoside Rh2 ; 16: ginsenoside
Rg2 ; 17: ginsenoside Rb2 ; 18: ginsenoside Rb3 ; 19: ginsenoside Rd; 20: chihusetsusaponin, and 21: ginsenoside Rh1 .

Table 3
UPLC/PDA peak areas of the compounds in CPC fractions 1–10 and the activities of these fractions.

Fraction UPLC/PDA peak area (mAu/s)

majonoside R2 notoginsenoside R6 notoginsenoside R1 ginsenoside Re ginsenoside Rg1 majoroside F4 activities

1 75679 95427 17919 443435 29324 – 59.74%

2 70437 78436 16425 657234 63470 – 60.74%
3 65520 47246 14721 789665 133341 – 62.10%
4 32548 22343 8033 659438 480428 – 59.05%
5 10742 10548 4742 543241 644074 – 57.58%
6 4345 5324 1203 173074 518457 – 51.70%
7 1034 1133 929 54079 128541 45.12%
8 – – – 25210 85545 986 33.88%
9 – – – 1631 5232 5441 26.45%
10 – – – 924 1325 37354 26.28%

bition rates of fractions 1–6 were above 50.00%. The activities of
these seven fractions were 59.74, 60.74, 62.10, 59.05, 57.58, 51.70,
and 45.12%, respectively. Using multilinear regression analysis, the
dependences of activities (y) of the seven fractions on tyrosinase
inhibition ability coefficients (x) can be well expressed as functions
of (logc1,1 ) · x1 + (logc1,2 ) · x2 + (logc1,3 ) · x3 + (logc1,n ) · xn = y1 , proven
by testing the least-square error. The functions corresponding to
different fractions were expressed as follows:
Fraction 1: log 75679 · x1 + log 95427 · x2 + log 17919 · x3 + log
443435 · x4 + log 29324 · x5 = 48.87%
Fraction 2: log 70437 · x1 + log 78436 · x2 + log 16425 · x3 + log
657234 · x4 + log 63470 · x5 = 51.31%
Fraction 3: log 65520 · x1 + log 47246 · x2 + log 14721 · x3 + log
789665 · x4 + log133341 · x5 = 53.53%
Fraction 4: log 32548 · x1 + log 22343 · x2 + log 8033 · x3 + log
659438 · x4 + log 480428 · x5 = 56.74%
Fraction 5: log 10742 · x1 + log 10548 · x2 + log 4742 · x3 + log
543241 · x4 + log 644074 · x5 = 56.89%
Fraction 6: log 4345 · x1 + log 5324 · x2 + log 1203 · x3 + log 173074
· x4 + log 518457· x5 = 54.59%
Fraction 7: log 1034 · x1 + log 1133 · x2 + log 929 · x3 + log 54079
· x4 + log 28541 · x5 = 43.54%
x1 , x2 , x3 , x4 , and x5 were calculated as 0.0043, 0.0067, 0.0093,
0.0142, and 0.0704, respectively.
This result indicates that the tyrosinase inhibition ability coef-
ficients of majonoside R2 , notoginsenoside R6 , notoginsenoside R1 ,
ginsenoside Re, and ginsenoside Rg1 were 0.0043, 0.0067, 0.0093,
0.0142, and 0.0704, respectively. However, the data mentioned
above were calculated coefficients, which should be translated into
real tyrosinase inhibition abilities.
Group 2 was composed of fractions 8–10 and contained ginseno-
sides Re, Rg1 , and majoroside F4 . The UPLC peak areas of the above
saponins in fractions 8–10 are listed in Table 3. The activities of the
three fractions were 45.30, 35.62, and 31.83%, respectively.
Fraction 8: log 25210 · x4 + log 85545 · x5 + log 986 · x6 = 45.30%
Fraction 9: log 1631 · x4 + log 5232 · x5 + log 5441· x6 = 35.62%
Fraction 10: log 924 · x4 + log 1325 · x5 + log 37354 · x6 = 31.83%
x4 , x5 , and x6 were calculated as 0.0136, 0.0731, and 0.0109,
respectively. The notoginsenosides R2 , R6 , and R1 contained in frac-
tions 1–7 were almost undetectable in fractions 8–10, even tiny
peaks were detected in fraction, but it should not be reflected
 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87
in the statistics. These results indicate that the tyrosinase inhibi-
tion ability coefficients of ginsenosides Re, Rg1 , and majoroside F4
were 0.0136, 0.0731, and 0.0109, respectively. For the first group,
the tyrosinase inhibition ability coefficients of ginsenosides Re and
Rg1 were calculated as 0.0142 and 0.0704, respectively, the abso-
lute values of the analysis error between two groups were 4.22
and 3.84%, and the calculated results were verified by two groups,
indicating their accuracy.
Groups 3–18 contained fractions 11–15, 16–20, 21–24, 25–30,
31–35, 46–48, 49–51, 52–56, 61–68, 72–73, 80–82, 83–97, 98–101,
105–112, 122–131, and 132–137, respectively. The compounds in
groups 3–11 are shown in Fig. 2 and Table S1. The UPLC peak areas of
compounds, activities of fractions 11–137, and functions related to
tyrosinase inhibition ability coefficients are shown in Tables S2–S6.
The tyrosinase inhibition ability coefficients of vina-ginsenoside
R10 , chikusetsusaponin III, ginsenosides Rf, Ra3 , notoginsenoside
R2 , chikusetsusaponin L5 , ginsenoside Rb1 , notoginsenoside Fa,
ginsenosides Rh2 , Rg2 , Rb2 , Rb3 , Rd, chihusetsusaponin, and gin-
senoside Rh1 were calculated and listed in Table S7. The tyrosinase
inhibition ability coefficients of the compounds mentioned above
were 0.0264, 0.0224, 0.0372, 0.0014, 0.0160, 0.0621, 0.0053, 0.0012,
0.0638, 0.0626, 0.0047, 0.0043, 0.0052, 0.0074, and 0.0758, respec-
tively. The inhibition rate of the ginsenoside Rh1 were obtained by
using the authentic standard, the results indicates that the inhibi-
tion rate of the ginsenoside Rh1 was 68.25% at a concentration of
30 ␮M. According to the ratio of tyrosinase inhibition ability coef-
ficients and the inhibition rate of ginsenoside Rh1 , the tyrosinase
inhibitions of majonoside R2 , notoginsenosides R6 , R1 , ginsenosides
Re, Rg1 , majoroside F4 , vina-ginsenoside R10 , chikusetsusaponin
III, ginsenosides Rf, Ra3 , notoginsenoside R2 , chikusetsusaponin L5 ,
ginsenoside Rb1 , notoginsenoside Fa, ginsenosides Rh2 , Rg2 , Rb2 ,
Rb3 , Rd, chihusetsusaponin, and ginsenoside Rh1 were calculated as
3.82, 5.99, 8.26, 12.63, 64.48, 9.65, 23.48, 19.94, 33.06, 1.26, 14.21,
55.26, 4.64, 1.08, 66.75, 56.33, 4.21, 3.87, 4.69, 6.65, and 68.25%,
respectively.
3.4. Online UPLC/PDA monitoring of CPC fractions
Coupling of CPC with online UPLC/PDA requires a connection
valve that delivers the CPC effluent to the UPLC column, which
is then eluted by the mobile phase. Injection volume is a crucial
parameter for UPLC analysis, since a high volume of CPC effluent
reduces the resolution of adjacent chromatographic peaks, whereas
a low volume decreases the signal intensity of target compounds,
resulting in relatively large analytical and calculation errors. Con-
sidering the factors mentioned above, the UPLC injection volume
was set at 2 ␮L.
The CPC effluent was injected into the UPLC column via a six-
port valve through an injection loop, which is easy to accomplish
for manual-injection UPLC. However, the latter has already stepped
down from the stage of history. More and more laboratories use
UPLC instruments with an autosampler. Thus, we designed an
injection mode for autosampler-equipped UPLC without modify-
ing or destroying the original instrument. The six-port valve was
placed between the UPLC column and the UPLC pump, and an ace-
tonitrile/water mixture (1:3 v/v) was placed in the injection vial
without the sample. After the CPC effluent filled the injection loop,
the CPC effluent was pumped from the injection loop into the UPLC
column for analysis.
3.5. Tyrosinase inhibition verification
The tyrosinase inhibition activities of standard compounds
including ginsenosides Re, Rg1 , Rf, Rb1 , Rg2 , Rb2 , Rd, Rh1 , and
notoginsenoside R2 were determined, with the results shown
in Table 4. The results indicate that ginsenosides Rh1 and Rg1
85
Table 4
Tyrosinase inhibition activities of standards.
Compounds Inbibition a,b IC50 a,c (␮M)
ginsenoside Re 10.81% >100.00
ginsenoside Rg1 55.33% 15.65
ginsenoside Rf 31.06% 73.27
ginsenoside Rb1 9.69% >100.00
ginsenoside Rg2 47.48% 32.27
ginsenoside Rb2 4.21% >100.00
ginsenoside Rd 10.06% >100.00
ginsenoside Rh1 61.25% 11.72
notoginsenoside R2 13.63% >100.00
a
The results were an average of three determinations.
b
Inhibition by 20 ␮M standard compounds.
c
Concentration required for 50% inhibition of the enzyme activity under the assay
conditions.
showed inhibitions of 61.25 and 55.33%, respectively, at 20 ␮M;
ginsenosides Rg2 , Rf, and notoginsenoside R2 showed smaller
inhibitory activities, the corresponding inhibition values being
47.48, 31.06, and 13.63%, respectively, and the IC50 values were
32.27, 73.27, and >100.00 ␮M, respectively. The inhibition rates of
other compounds were very small, close to or less than 10% at a
concentration of 20 ␮M. The order of tyrosinase inhibition rates
was as follows: ginsenoside Rh1 > Rg1 > Rg2 > Rf > notoginsenoside
R2 > Re > ginsenoside Rd > Rb1 > Rb2 . The results were consistent
with the calculated data. Since the activities of ginsenoside Rh1 , Rg1 ,
Rg2 , and Rf were relatively high, other verifications were under-
taken. First, correlations between UPLC peak areas (mU/s) and
concentrations of ginsenosides Rg1 , Rf, Rg2 , and Rh1 were calcu-
lated, showing good linear relationships, with the errors R2 all being
less than 0.9986, and the calculated error r being less than 0.9993
(Fig. 3). Second, the correlations between tyrosinase inhibition and
log concentration, and those between inhibition and log UPLC peak
areas (mU/s) of the compounds were calculated. The results indi-
cates that the above correlations were relatively linear, with the
error R2 for the correlation of inhibition and log UPLC peak area,
and the correlation of inhibition and log concentrations being less
than 0.9925 and 0.9935, respectively. These results provide a basis
for fitting UPLC peak areas and activities of the CPC fractions by
multivariate linear functions.
The protopanoxatriol ginsenosides (ginsenosides Rh1 , Rg2 , and
Rg1 ) were more potent tyrosinase inhibitors than the protopanoxa-
diol ones (ginsenosides Rb1 , Rb2 , and Rd), and the activities
of low-molecular-weight ginsenosides were much higher than
those of the high-molecular-weight ones. This phenomenon may
be attributed to the lower number of glycosylation moieties in
low-molecular-weight ginsenosides compared to high-molecular-
weight ones, as shown in Table 4. A possible reason for the
decreased inhibitory activity observed are the steric hindrance
effects of the bulky glycoside moiety, which could hinder the
approach of large molecules to the active site of the enzyme [29,40].
In addition, the tyrosinase inhibition ability coefficients were
based on chromatographic peak areas and mathematical calcula-
tions. The data represents the tyrosinase inhibition abilities; if the
peak areas are converted to concentrations, the true inhibition data
and IC50 values can be obtained.
4. Conclusions
Automated coupling of ASE, CPC, and UPLC/PDA was devel-
oped and applied for the first time to the extraction, fractionation,
and simultaneous composition analysis of complex mixtures. This
coupled system allowed us to quickly obtain the first extrac-
tion and perform the fractionation and identification of saponins
from P. bipinnatifidus Seem. on a semi-preparative or analytical
scale without preliminary treatment. Compared with the offline
86 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87

Fig. 3. Correlations between inhibition and log concentration (␮M) and those between inhibition and log UPLC peak areas (mU/s) of ginsenosides Rg1 , Rf, Rg2 , and Rh1 .
Correlations between UPLC peak areas (mU/s) and concentrations of ginsenosides Rg1 , Rf, Rg2 , and Rh1 .
technique, ASE/CPC/UPLC/PDA avoids many post-extraction and
post-CPC steps such as filtration, tube transfer, evaporation, and
dissolution in appropriate solvent and simplifies the pooling of frac-
tions. Consequently, this system decreases the analysis time and
the risk of mistakes and chemical degradation by integrating three
steps (i.e., fractionation, analysis, and identification) into one. The
results obtained for complete automation are highly repeatable,
confirming that this technique can be routinely used in any labora-
tory. Furthermore, the process is highly integrated and automated,
showing great potential for industrial application.
The traditional and/or general applications of CPC were changed,
since we fractionated the CPC effluent instead of isolating the pure
compounds. Analysis of CPC fractions by online UPLC/PDA and
biological activity evaluation helped calculate the activities of com-
pounds therein, overcoming the burden of isolating large numbers
of pure compounds.
With the help of the methods established in this article,
tyrosinase inhibition abilities of compounds contained in P.
bipinnatifidus Seem. were obtained, the most active one being
ginsenoside Rh1 , with a tyrosinase inhibition coefficient of 0.0758
and an inhibition rate of 68.28%. The order of compound inhibition
ability, including ginsenoside Rh1 , as follows: ginsenoside Rh1
(inhibition rate of 68.25%) > ginsenoside Rh2 (66.75%) > ginsenoside
Rg1 (64.48%) > ginsenoside Rg2 (56.33%) > chikusetsusaponin L5
(55.26%) > ginsenoside Rf (33.06%) > vina-ginsenoside R10
(23.48%) > chikusetsusaponin III (19.94%) > notoginsenoside
R2 (14.21%) > ginsenoside Re (12.63%) > majoroside F4
(9.65%) > notoginsenoside R1 (8.26%) > chihusetsusaponin
(6.65%) > notoginsenoside R6 (5.99%) > ginsenoside Rd
(4.69%) > ginsenoside Rb1 (4.64%) > ginsenoside Rb2
(4.21%) > ginsenoside Rb3 (3.87%) > majonoside R2
(3.82%) > ginsenoside Ra3 (1.26%) > notoginsenoside Fa (1.08%). The
calculated inhibition rates of ginsenosides Rh1 , Rh2 , Rg1 , Rg2 , and
chikusetsusaponin L5 are all above 50.00%, showing potential for
further development. The tested inhibition rates of ginsenosides
Rh1 , Rg1 , and Rg2 at a concentration of 20 ␮M were determined as
61.25, 55.33, and 47.48%, respectively, supporting the calculated
results.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China (Nos. 31370374 and 31500279), the Nat-
ural Science Foundation of Jilin Province (Nos. 20150520140JH,
20140311032YY), the China Postdoctoral Science Foundation
(2016M590267), and the Natural Science Foundation of Changchun
Normal University ([2015]008, [2015]009).
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.jchromb.2016.
10.043.
References
[1] Y. Cai, Q. Luo, M. Sun, H. Corke, Life Sci. 74 (2004) 2157.
[2] I. Raskin, D.M. Ribnicky, S. Komarnytsky, N. Ilic, A. Poulev, N. Borisjuk, A.
Brinker, D.A. Moreno, C. Ripoll, N. Yakoby, J.M. O’Neal, T. Cornwell, I. Pastor, B.
Fridlender, Trends Biotechnol. 20 (2002) 522.
[3] M.F. Balandrin, J.A. Klocke, E.S. Wurtele, W.H. Bollinger, Science 228 (1985)
1154.
[4] X. Huang, L. Kong, X. Li, X. Chen, M. Guo, H. Zou, J. Chromatogr. B 812 (2004)
71.
[5] S. Sasidharan, Y. Chen, D. Saravanan, K.M. Sundram, L. Yoga Latha, Afr. J.
Tradit. Complement. Altern. Med. 8 (2011) 1.
[6] C.W. Huie, Anal. Bioanal. Chem. 373 (2002) 23.
[7] Y. Zhang, L. Guo, C. Liu, Z. Fu, L. Cong, Y. Qi, D. Li, S. Li, J. Wang, J. Chromatogr. B
935 (2013) 16.
[8] Y. Zhang, L. Guo, C. Liu, Y. Qi, D. Li, S. Li, Sep. Purif. Technol. 116 (2013) 240.
[9] N. Angelova, H.W. Kong, R. van der Heijden, S.Y. Yang, Y.H. Choi, H.K. Kim, M.
Wang, T. Hankemeier, J. van der Greef, G. Xu, R. Verpoorte, Phytochem. Anal.
19 (2008) 2.
[10] Y.H. Chuang, Y. Zhang, W. Zhang, S.A. Boyd, H. Li, J. Chromatogr. A 1404 (2015)
1.
[11] A.P. Foucault, J. Chromatogr. A 906 (2001) 365.
[12] Y. Zhang, C. Liu, J. Li, Y. Qi, Y. Li, S. Li, Ultrason. Sonochem. 26 (2015) 111.
[13] L.W. Qi, C.Z. Wang, C.S. Yuan, Nat. Prod. Rep. 28 (2011) 467.
[14] T. Michel, E. Destandau, L. Fougère, C. Elfakir, Anal. Bioanal. Chem. 404 (2012)
2963.
[15] V. García, J. Päkkilä, H. Ojamo, E. Muurinen, R.L. Keiski, Renew. Sustain. Energy
Rev. 15 (2011) 964.
[16] A.D. LaFleur, K.A. Schug, Anal. Chim. Acta 696 (2011) 6.
[17] S.B. Bankar, S.A. Survase, H. Ojamo, T. Granström, Bioresour. Technol. 140
(2013) 269.
[18] B.D. Thach, L.T. Linh, N.T. Van, T.T. Ben, M. Truong, N.H. Ho, J. Dev. Biol. Tissue
Eng. 6 (2014) 1.
[19] H.T. Nguyen, H.Q. Tran, T.T. Nguyen, V.M. Chau, K.A. Bui, Q.L. Pham, M.C.
Nguyen, Y.H. Kim, Chem. Pharm. Bull. 59 (2011) 1417.
[20] J.S. Choi, K.S. Chun, J. Kundu, J.K. Kundu, Int. J. Mol. Med. 32 (2013) 1227.
[21] T.G. Lim, C.C. Lee, Z. Dong, K.W. Lee, Arch. Dermatol. Res. 307 (2015) 397.
[22] L.P. Christensen, Nutr. Res. 55 (2009) 1.
[23] C.Z. Wang, S.R. Mehendale, C.S. Yuan, Am. J. Chin. Med. 35 (2007) 543.
[24] C.Z. Wang, E. Mcentee, S. Wicks, J.A. Wu, C.S. Yuan, J. Nat. Med. 60 (2006) 97.
[25] A.A.E.M. Abdurahman, K.Y. Musa, U.A. Zezi, Planta Med. 72 (2006) 1015.
 Y. Zhang et al. / J. Chromatogr. B 1039 (2016) 79–87 87

[26] L.K. Han, Y.N. Zheng, M. Yoshikawa, H. Okuda, Y. Kimura, BMC Complement.
Altern. Med. 5 (2005) 1.
[27] D.S. Kumar, D. Banji, A. Harani, J. Pharm. Res. 4 (2011) 597–600.
[28] S. Liu, J. Yan, J. Xing, F. Song, Z. Liu, S. Liu, J. Pharm. Biomed. Anal. 59 (2012) 96.
[29] Z.Z. Yang, Y.F. Zhang, L.J. Sun, Y. Wang, X.M. Gao, Y.Y. Cheng, Anal. Chim. Acta
719 (2012) 87.
[30] K.A. Mortier, W.E. Lambert, J. Chromatogr. A 1108 (2006) 195.
[31] S. Taylor, A. Harker, J. Pharm. Biomed. Anal. 41 (2006) 299.
[32] M. Okomo-Adhiambo, T.G. Sheu, L.V. Gubareva, Influenza Other Respir. 7
(2013) 44.
[33] H. Marjuki, V.P. Mishin, K. Sleeman, M. Okomo-Adhiambo, T.G. Sheu, L. Guo, X.
Xu, L.V. Gubareva, Antimicrob. Agents Chemother. 57 (2013) 5209.
[34] R.Y. Cao, J.H. Xiao, B. Cao, S. Li, Y. Kumaki, W. Zhong, Antiviral Chem.
Chemother. 23 (2014) 237.
[35] Y.W. Ha, S.S. Lim, I.J. Ha, Y.C. Na, J.J. Seo, H. Shin, S.H. Son, Y.S. Kim, J.
Chromatogr. A 1151 (2007) 37.
[36] Y. Ito, J. Chromatogr. A 1065 (2005) 145.
[37] Z. Yang, Y. Zhang, L. Sun, Y. Wang, X. Gao, Y. Cheng, Anal. Chim. Acta 719
(2012) 87.
[38] Y.Y. Liu, J.B. Li, J.M. He, Z. Abliz, J. Qu, S.S. Yu, S.G. Ma, J. Liu, D. Du, Rapid
Commun. Mass Spectrom. 23 (2009) 667.
[39] S. Yahara, R. Kasai, O. Tanaka, Chem. Pharm. Bull. 25 (1977) 2041.
[40] Y.P. Lin, F.L. Hsu, C.S. Chen, J.W. Chern, M.H. Lee, Phytochemistry 68 (2007)
1189.