J. Sep. Sci.

2014, 37, 2331–2339 2331

Saı̈da Chideh1,2 Research Article

Serge Pilard3
Jacques Attoumbré4
Robert Saguez1 5-O-Caffeoylshikimic acid from Solanum
Alshaimaa Hassan-Abdallah5
Dominique Cailleu3 somalense leaves: Advantage of centrifugal
Anne Wadouachi6
Sylvie Baltora-Rosset1
partition chromatography over conventional
1 EDYSAN
column chromatography
FRE 3498
CNRS-Université de Picardie
Jules Verne, UFR de Pharmacie, Solanum somalense leaves, used in Djibouti for their medicinal properties, were extracted by
Amiens Cedex, France MeOH. Because of the high polyphenol and flavonoid contents of the extract, respectively,
2 Centre de Recherche, Université
determined at 80.80 ± 2.13 mg gallic acid equivalent/g dry weight and 24.4 ± 1.01 mg
de Djibouti, Avenue Georges
quercetin equivalent/g dry weight, the isolation and purification of the main polyphenols
Clémenceau, Djibouti
3 Plate-Forme Analytique, UFR were carried out by silica gel column chromatography and centrifugal partition chromatog-
des Sciences, Bâtiment raphy. Column chromatography led to 11 enriched fractions requiring further purification,
Serres-Transfert, Amiens Cedex, while centrifugal partition chromatography allowed the easy recovery of the main compound
France
4 S.I.P.R.E – Comité Nord, of the extract. In a solvent system composed of CHCl3 /MeOH/H2 O (9.5:10:5), 21.8 mg of
Achicourt, France this compound at 97% purity was obtained leading to a yield of 2.63%. Its structure was
5 Institut de Recherches
established as 5-O-caffeoylshikimic acid by mass spectrometry and NMR spectroscopy. This
Medicinales, CERD, Djibouti work shows that S. somalense leaves contain very high level of 5-O-caffeoylshikimic acid
6 LG2A FRE 3517
CNRS-Université de Picardie (0.74% dry weight), making it a potential source of production of this secondary metabolite
Jules Verne, Institut de Chimie that is not commonly found in nature but could be partly responsible of the medicinal
de Picardie FR 3085, Amiens properties of S. somalense leaves.
Cedex, France

Keywords: Bioautography / Caffeoylshikimic acids / Dactylifric acid / Polyphenols

Received March 3, 2014 / Solanum somalense
Revised June 11, 2014 DOI 10.1002/jssc.201400226
Accepted June 12, 2014

1 Introduction Bioactive known compounds of plants are divided into

Investigating the scientific basis for the use of medicinal
plants is of great interest for several reasons: (i) plants re-
main the primary source of traditional medicine, (ii) a marked
increase in the use of herbal medicines or botanical dietary
supplements is observed in Western medicine, and (iii) plant-
derived drugs are the main sources for finding lead com-
pound structures that might reach the drug market [1, 2].
Thousands of traditional herbal medicines have been widely
used and studied but Djiboutian ones have not received much
attention up to now despite their significant contribution
to the management of public healthcare in the country [3].
Among the different medicinal plants used in the Randa re-
gion in Djibouti, Solanum somalense has various medicinal
uses but, as far as we know, has not been studied from a
phytochemical and pharmacological point of view [3].
Correspondence: Dr. Sylvie Baltora-Rosset, EDYSAN FRE 3498
CNRS-Université de Picardie Jules Verne, UFR de Pharmacie, 1
rue des Louvels, 80037 Amiens Cedex, France
E-mail: sylvie.baltora-rosset@u-picardie.fr
Fax: +33-03-2282-7469
Abbreviations: CPC, centrifugal partition chromatography;
DPPH, 2,2-diphenyl-1-picrylhydrazyl; DW, dry weight; GAE,
gallic acid equivalent; QE, quercetin equivalent
three main categories: (i) terpenes and terpenoids (approx-
imately 25 000 types), (ii) alkaloids (approximately 12 000
types), and (iii) phenolic compounds (approximately 8000
types) [4]. Given the difficulty of a comprehensive analysis
of plant extracts, studies usually focus on a limited range of
compound classes often highlighted by a preliminary rapid
biological screening of the total extract. Criteria of chemotax-
onomy and uses in traditional medicine may also guide the
choice of the class of compounds that will be targeted [4].
The relationship between various diseases such as can-
cer, hypertension, diabetes, cardiovascular, neurodegenera-
tive diseases, and oxidative damages has been extensively in-
vestigated and established [5, 6]. At the same time, the search
for antioxidants from natural sources to prevent these oxida-
tive damages is currently under development. Among sec-
ondary metabolites, polyphenols generally exhibit good an-
tioxidant properties and are thus often sought in medicinal
plants but may cause false-positive results in enzymatic and
cellular screening procedures due to their protein-binding
ability and their changes in cellular redox potential [7,8]. Var-
ious analytical methods have been reported for the analysis
and purification of bioactive natural compounds [4, 7, 9] and
for the most abundant polyphenols in plants, phenolic acids,
and flavonoids [10]. Numerous articles on their extraction,
separation, and detection methods have been published over
the past two decades [11]. SPE methods are currently used


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
2332 S. Chideh et al.
to recover prefractionated samples or to remove polyphenols
from plant extracts because SPE consists of simple elution
and concentration via polymeric resins. In practice, SPE of-
ten suffers from low resolution and efficiency [7]. However,
SPE is frequently used for cleanup procedures, mainly be-
cause it offers the possibility of combining on-line extrac-
tion with chromatographic methods such as HPLC [11]. The
main methods described for the determination of phenolic
acids and flavonoids are chromatographic techniques such
as silica gel or sephadex column chromatography, TLC, GC,
HPLC, or LC–ESI-MS [4, 11]. Because the chemical struc-
tures of polyphenols have various degrees and patterns of
hydroxylation, methylation or glycosylation, their analysis by
GC requires a preliminary chemical derivatization. Actually,
their fractionation from natural sources is usually performed
using a combination of methods, mainly column chromatog-
raphy followed by HPLC [12]. To overcome the use of multiple
steps, a new method, countercurrent separation, based solely
on differences in the solubility of natural products was de-
veloped in the 1970s [3, 13, 14]. CCC is a chromatographic
technique that has some advantages when compared with LC
techniques: (i) no nonspecific adsorption to a solid support,
(ii) higher selectivity, (iii) higher sample loading capacity,
(iv) reduction of solvent quantity, and (v) shorter separation
time. Therefore, CCC has been successfully applied to the
analysis and separation of various natural products including
polyphenols [9, 11–13, 15].
Caffeoylshikimic acids are characteristic phenolic com-
pounds mainly found in the Palmae family [16]. Up to now,
the largest industrial use of shikimic acids has been the pro-
duction of Tamiflu R
or other antiviral drugs for use against
avian flu, and the demand for shikimic acid and its deriva-
tives is expected to increase dramatically in the event of a
pandemic flu outbreak. Zeng et al. [17] also demonstrated the
antiproliferative activity of 5-O-caffeoylshikimic acid against
different cancer cell lines. Very few plant sources of these
acids have been described. Elaeis guineensis (oil palm) and
Phoenix dactylifera (date palm) are the only sources where
caffeoyl shikimic acids are the major polyphenols [16,18–20].
5-O-Caffeoylshikimic acid was also obtained in small quanti-
ties from Vaccinium corymbosum (10 mg for 32 g of extract)
together with 21 phenolics and showed the potential for nu-
traceutical applications [21]. The compound was also isolated
from Livistona chinensis through a multistep procedure lead-
ing to a very low yield: 12 mg for 90.21 g of crude extract.
The aim of this work was to develop an efficient bioassay-
guided method for the fractionation of the methanolic extract
of S. somalense. The preliminary screening of the antioxidant
properties of the extract, as well TLC analysis with selective
polyphenol developers, suggested that the methanolic extract
had a high polyphenol content. As described above, a review
of the literature indicates that no single method is regarded as
standard for extracting polyphenols from plants. In addition,
to our knowledge, for caffeoylshikimic acids, only multistep
methods including column chromatography and HPLC are
described. Therefore, it seemed interesting to develop a new
method for isolating these types of phenolic components by

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2014, 37, 2331–2339
the use of centrifugal partition chromatography (CPC). As
conventional chromatographic separation on silica gel is con-
sidered to be one of the reference methods for comparing the
success of newly developed methodologies, the two methods
were carried out in parallel. Our objective was fully achieved
and led to the successful preparative isolation of highly pure
major phenolic component from the S. somalense leaves. 5-
O-Caffeoylshikimic acid was identified by ESI-MS and NMR
spectroscopy and will be used for further biological studies.
This paper is the first report on the separation of bioactive
compounds of S. somalense by CPC.
2 Materials and methods
2.1 Materials
Solanum somalense leaves were collected from Tadjourah Dis-
trict of Randa Region in north Djibouti. Plant identification
was carried out at the National Herbarium (ETH), Addis
Ababa University (AAU) and voucher specimens were de-
posited in the Herbarium at Djibouti. All organic solvents
were of analytical grade and purchased from VWR. Reagents
were purchased from Sigma–Aldrich.
2.2 Biological evaluation
Determination of total polyphenols. Total phenolic contents
were determined according to the literature [19]. A total of
0.50 mL of the sample diluted in MeOH was added to 2.5 mL
of 1:10 diluted Folin–Ciocalteu reagent. After 4 min, 2 mL
of saturated sodium carbonate solution (about 75 g/L) was
added. After 2 h of incubation at room temperature, the ab-
sorbance of the reaction mixture was measured at 760 nm.
Gallic acid was used as a reference standard and the re-
sults were expressed as milligram gallic acid equivalent (mg
GAE)/(g dry weight (DW)) of plant material. Gallic acid con-
centrations ranging from 0 to 100 ␮g/mL were prepared, and
the standard calibration curve was obtained using a linear fit
(r2 = 0.9972). The samples were analyzed in triplicate.
Determination of total flavonoids. Total flavonoids were an-
alyzed according to the aluminum chloride method described
in Ref. [8]. A total of 0.5 mL of each sample and 300 ␮L of
NaNO2 (1:20 w/v) were pipetted into a test tube. The con-
tents were vortexed for 10 s and left at room temperature for
5 min. Three hundred microliters of AlCl3 (1:10 w/v), 2 mL
of 1 M NaOH, and 1.9 mL of distilled water were then added
to the mixture. After 10 s of vortexing, the absorbance for
each sample was measured at 510 nm. Quercetin concentra-
tions ranging from 0 to 120 ␮g/mL were prepared, and the
standard calibration curve was obtained using a linear fit (r2
= 0.9831). Quercetin was used as a reference standard and
the results were expressed as (mg QE)/(g DW) (where QE
is quercetin equivalent) of plant material. The samples were
analyzed in triplicate.
www.jss-journal.com
J. Sep. Sci. 2014, 37, 2331–2339
2.3 Apparatus
The CPC instrument used in this study is a SPOT CPC 100
Light (Armen Instrument) fitted with a rotor of ten circular
partition disks (1000 partition cells: 0.1 mL per cell; total col-
umn capacity of 100 mL). The rotation speed can be chosen
from 0 to 4000 rpm. The effluent was monitored by a Lash
06 DAD detector (ECOM, Prague) equipped with a prepara-
tive flow cell operating at 254 nm and collected by a LS 5600
(Armen) fraction collector. The HPLC used was a Shimadzu
HPLC System including a LC-20AT pump and SPD-M20A
diode-array detector. LC–ESI-MS spectra were obtained on a
UFLC Prominence (Shimadzu) system coupled with a Q-TOF
Ultima Global high-resolution hybrid quadrupole TOF instru-
ment (Waters-Micromass), equipped with a pneumatically
assisted electrospray ionization source (Z-spray) and an addi-
tional sprayer (lock spray) for the reference compound, allow-
ing accurate mass measurements of ions and their elemental
compositions determination. NMR spectra were recorded at
300 K on a Bruker (Wissembourg, France) DRX-500 spec-
trometer equipped with a broadband inverse probe. DMSO-
d6 was used as the solvent. Proton resonance assignments
and structural elucidation were performed by 1D (1 H and 13 C
NMR) and 2D homo- and heteronuclear experiments (corre-
lated spectroscopy, heteronuclear single quantum coherence,
and heteronuclear multiple-bond correlation).
2.4 Preparation of crude extract
Fifty grams of dried leaves of S. somalense were powdered
and treated three times with 500 mL of cyclohexane. After
filtration through Whatman filter paper, the plant material
was subjected to three extractions with AcOEt (500 mL each
time, room temperature, 24 h). The same procedure was ap-
plied with MeOH and with H2 O. The extracts from the same
solvent were combined and the solvent was removed under
the reduced pressure to give the crude extracts.
2.5 Detection by TLC, HPLC, and LC–MS
2.5.1 TLC
Fractionation and separation were monitored by TLC car-
ried out on Silica Gel 60 F254 (Merck) plates developed
with EtOAc/HCOOH/HOAc/H2 O (100:11:11:20) visualized
by UV light (254 and 365 nm) and/or with vanillin-H2 SO4
reagent, Neu reagent (1% diphenylboric acid ethanolamine
complex in EtOH) and 2,2-diphenyl-1-picrylhydrazyl (DPPH
solution, 2 mg/mL MeOH).
2.5.2 HPLC
HPLC analyses of the crude extract and of the CPC peak
fractions were conducted at 254 and 365 nm on a 250 ×
4.6 mm, 5 ␮m, Prevail RP C18 column (Grace) using a linear

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 2333
binary gradient of H2 O (solvent A) and MeCN (solvent B) both
containing 0.1% v/v HCOOH, with a flow rate of 1 mL/min:
5–95% B to 95–5% B in 45 min. Twenty microliters was used
for injection, which was repeated three times.
2.5.3 LC–ESI-MS
The crude extract and the purified fractions were loaded
(2 ␮L) on a 100 × 2 mm, 2.7 ␮m Nucleoshell RP18
column (Macherey Nagel) and maintained at 30⬚C. The elu-
tion was performed using a 0.25 mL/min mobile phase gra-
dient programmed from water (A), acetonitrile (B), both con-
taining 0.1% v/v HCOOH, as follows (A/B): 95:5 (t = 0 min),
95:5 (t = 5 min), 5:95 (t = 30 min), 5:95 (t = 40 min), 95:5 (t
= 45 min), and 95:5 (t = 50 min). The effluent was flow-split
via a polyether ether ketone tee with one-third of the flow
directed toward the electrospray source of the Q-TOF and
the residual two-third directed toward an UV detector (Shi-
madzu SPD-20A) set to 254 nm. ESI-MS data were recorded
in the positive- and negative-ion modes with capillary voltage
of ±2.5 kV and cone voltage of 50 V. The source and desolva-
tion temperatures were 120 and 250⬚C, respectively. Nitrogen
was used as a drying and nebulizing gas at flow rates of 450
and 100 L/h, respectively. Scanning was performed in the
range 50–1550 Da at a scan rate of 1 s/scan and spectra were
collected in the profile mode at a resolution of 10 000 (full
width at half maximum). Lock mass correction, using appro-
priate cluster ions of an orthophosphoric acid solution (0.1%
in H2 O/CH3 CN 50:50, v/v), was applied for accurate mass
measurements. For MS/MS experiments, argon was used as
collision gas and the collision energy was set to 20 V. Data
acquisition and processing were performed with MassLynx
4.0 SP4 software.
2.6 Separation methods
2.6.1 Silica gel column chromatography separation
Half of the methanolic crude extract (3.9 g) was subjected to
silica gel (Merck 70–230 ␮m) column chromatography eluted
with AcOEt/MeOH/H2 O (100:0:0, 90:10:1, 80:20:1, 70:30:1,
60:40:1, 50:50:1, and 40:60:1). This separation resulted in 64
fractions that were checked for their composition by TLC with
four types of visualizers: UV254 , vanillin sulfuric, Neu reagent,
and DPPH test. This combination of visualization allowed the
gathering of the initial fractions into 11 fractions (F1–F11),
which were weighed and submitted to further analyses.
2.6.2 CPC separation
2.6.2.1 Selection of the two-phase solvent system
The solvent system was selected according to the distribution
constant KD of the six main peaks of the unknown com-
pounds (X) (called A–F) visualized on HPLC chromatogram
of the crude methanolic extract at 365 nm. The KD value was
www.jss-journal.com
2334 S. Chideh et al.
determined by HPLC analysis in the same conditions.
Methanolic crude extract was dissolved in the tested solvent
system and vortexed for 30 s. After separation and evapora-
tion under reduced pressure, the residue of each layer was
dissolved in 500 ␮L of methanol for HPLC analysis. The KD
values were calculated according to the ratio: concentration of
compound X in the stationary phase/concentration of com-
pound X in the mobile phase [20].
2.6.2.2 Collection and analysis of fractions
The solvent system used for separation was chloro-
form/methanol/water in the ratio 9.5:10:5 v/v/v. The biphasic
system was prepared just before use by thoroughly mixing
volumes of solvent in the above ratio. After the equilibration
was established, the stationary phase (upper phase in the de-
scending mode) was pumped into the column at a flow rate of
30 mL/min while the apparatus was run at 500 rpm accord-
ing to the equilibration mode of the apparatus. After injection
of the sample (0.83 g of crude extract in 10 mL of a mixture
50:50 of the two phases), the mobile phase was perfused at
2500 rpm at a flow rate of 4 mL/min for 60 min and then to
8 mL/min under 35 b during the run. The eluent was moni-
tored at 254 nm and 36 fractions (fi ) of 15 mL were collected
and analyzed by HPLC. The volume of the stationary phase
displaced by the mobile phase was measured and used to de-
termine VM . The stationary phase volume (VS ) was calculated
according to the relationship VC = VM + VS since the column
volume (VC ) is known. The stationary phase retention Sf was
expressed as VS /VC (70% in the experiment). The selectivity
(separation factor between two compounds X and X ) was
calculated as follows: ␣ = KD(X) /KD(X ) [20].
3 Results and discussion
3.1 Biological evaluation of the extract
Leaves of S. somalense, which are mainly used for the prepara-
tion of traditional medicines, were extracted with solvents of
different polarity: cyclohexane, AcOEt, MeOH, and water. The
residues obtained by sequential extraction were weighed and
analyzed by TLC (data not shown). The extraction yields were
1.8, 1.3, 14, and 20%, respectively. The methanolic and water
extracts had the highest content of secondary metabolites and
showed quite similar TLC profiles, but the water residue was
quickly subjected to microbial contamination, which is one

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2014, 37, 2331–2339
Figure 1. TLC plates developed
with EtOAc/HCOOH/HOAc/H2 O (100:
11:11:20) and stained with (A) Neu
reagent visualized at 365 nm, (B) DPPH
solution in MeOH. Dots (100 ␮g) are
of (a) methanolic crude extract of S.
somalense (b) F1 (c) F5 (d) F8 (e) F10
(f) F11 (silica gel column fractions) (g)
compound B isolated by CPC.
of the main problems with the use of water plant extracts or
decoctions [21]. Under these conditions, only the methano-
lic extract was further analyzed by TLC in different detection
conditions. The main phytochemicals were detected by UV,
vanillin sulfuric acid, and Neu reagent while the antioxidant
activity was evaluated using DPPH radical scavenging assay
according to Zhang et al. [8]. The results of revelation by Neu
reagent and DPPH (Fig. 1A and B, respectively) showed that
polyphenols were present in large quantities in the methano-
lic extract from S. somalense leaves and also exhibited a high
antioxidant total activity due to a DPPH radical scavenging
capacity of almost all the compounds present in the extract.
Therefore, the total contents of polyphenol and flavonoid
were quantitatively evaluated and assessed at 80.80 ± 2.13 mg
GAE/g DW and 24.4 ± 1.01 mg QE/g DW, respectively. In
their study of the antioxidant properties of methanol extracts
of 45 whole medicinal herbs, Li et al. [22] showed that their
polyphenols contents could range from 1 to 52 mg GAE/g
DW. In this context, the value obtained for S. somalense is
very high. Furthermore, it was shown by the same authors
and others [8, 22, 23] that a high correlation existed between
high levels of polyphenols and the antioxidant capacities of
the studied extracts. Finally, the effects of plants on health
are often attributed to their polyphenolic components [10].
3.2 Analytical study of the extract
While keeping in mind that it is necessary to perform more
than one type of antioxidant capacity measurement to take
into account the various mechanisms of antioxidant actions
of plant extracts [24], we decided to study the major polyphe-
nols present in the extract to evaluate their potential health-
promoting benefits identified by the ethnological surveys
in Djibouti [3]. The analysis of the crude extract by HPLC
and LC–ESI-MS (Fig. 2A) indicated the presence of six main
peaks, which were characterized by their retention times (Rt),
their UV absorption maxima, and their molecular weight
(Table 1).
3.3 Separation and determination of polyphenols
To separate and identify these major polyphenols, a conven-
tional method such as activity-guided fractionation on silica
www.jss-journal.com
J. Sep. Sci. 2014, 37, 2331–2339 Liquid Chromatography 2335

Figure 2. (A) LC chromatogram at 365 nm of methanolic crude extract of S. somalense, (B) LC chromatogram at 365 nm of fraction F1
obtained by silica gel column chromatography, and (C) LC chromatogram at 365 nm of fraction f23 obtained by CPC separation. Conditions:
Prevail RP C18 column (250 × 4.6 mm, 5 ␮m), using a linear binary gradient of H2 O (solvent A) and MeCN (solvent B) both containing 0.1%
v/v HCOOH, with a flow rate of 1 mL/min: 5–95% B to 95–5% B in 45 min.

gel column chromatography was first performed and allowed
us to obtain 11 fractions Fi (Fig. 3). As shown by TLC (Fig. 1A),
gel column chromatography did not allow an efficient sepa-
ration of polyphenols: only the major compound of the crude
extract (peak B) was obtained with a yield of 1.08% but its pu-
rity was not sufficient (87%) (Figs. 1A and 2B) to consider fur-
ther biological evaluations because the minor compound in
the fraction also showed an antioxidant activity. Nevertheless,
this step was very useful to roughly characterize the six main
components of the extract detectable by HPLC at 365 nm and
to identify the structure of the major compound B (Table 1).
The LC–ESI-MS spectra (Fig. 4A and B) revealed [M+H]+ and
[M−H]− ions at m/z 337.09 and 335.07, respectively. A molec-
ular formula of C16 H16 O8 was determined using accurate
mass measurements, [M+H]+ : found 337.0935 for 337.0923
calculated and [M−H]− : found 335.0777 for 335.0767 calcu-
lated. In a positive mode (Fig. 4A), an abundant ion at m/z
163.03 (C9 H7 O3 ) seems to indicate the presence of a caffeoyl
moiety. Moreover UV spectrum of B showed absorption max-
ima at 211 and 316 nm suggesting a phenolic acid structure.


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
2336 S. Chideh et al. J. Sep. Sci. 2014, 37, 2331–2339

Table 1. LC retention time, ␭max (UV nm), molecular weight, and TLC retention factor of compounds A–F

Peak Retention timea) UV ␭max (nm) Molecular weight Retention factorb)

A 25.9 210, 235, 317 354 0.53

B 29.1 211, 316 336 0.90
C 35.4 208, 255, 354 610 0.43
D 36.3 197, 255, 282, 355 594 nd
E 38.1 199, 250, 346 464 nd
F 38.8 199, 265, 347 594 nd

nd, not determined.

a) Conditions: analysis at 365 nm on a 250 × 4.6 mm, 5 ␮m, Prevail RP C18 column using a linear binary gradient of H2 O (solvent A) and
MeCN (solvent B) both containing 0.1% v/v HCOOH, with a flow rate of 1 mL/min: 5–95% B to 95–5% B in 45 min.
b) Conditions: Silica Gel 60 F254 (Merck) plates developed with EtOAc/HCOOH/HOAc/H2 O (100:11:11:20).

Solanum somalense leaves
50.0 g
Methanolic extract
7.8 g
Silica-gel column Centrifugal Partition
Chromatography Chromatography
3.9 g 0.83 g
11 fractions 38 fractions
Total fraction weight: 2.14 g f19 (7.1 mg 86%) f20 (7.2 mg 91%)
F1 weight: 42.4 mg f21-f23 (21.8 mg 97%)
Purity 87 % f24 (5.8 mg 85%)
Figure 3. Processing flowchart for the preparative extraction of
the methanolic crude extract of S. somalense leaves by conven-
tional column chromatography and by CPC. The purity of each
fraction was determined by analytical HPLC. Conditions: Prevail
RP C18 column (250 × 4.6 mm, 5 ␮m), using a linear binary gra-
dient of H2 O (solvent A) and MeCN (solvent B) both containing
0.1% v/v HCOOH, with a flow rate of 1 mL/min: 5–95% B to 95–5%
B in 45 min.
The results of combined NMR experiments established the
structure of B unambiguously as 5-O-caffeoylshikimic acid
as described by Fukuoka [25] (data not shown). This was
confirmed by MS/MS of the [M−H]− ion (Fig. 4C) showing
the caffeoyl fragment (m/z 179.04, C9 H7 O4 ) and its char-
acteristic loss of H2 O (m/z 161.03, C9 H5 O3 ) and CO2 (m/z
135.05, C8 H7 O2 ) [26]. Solanum species contain abundant caf-
feic acid derivatives that have implications in human health.
Numerous studies have shown that efficacy and potency of
these derivatives are very much dependant on their struc-
ture and this strong structure–function relationship make
them good candidates for the development of new functional
foods or pharmaceuticals [27]. Caffeic acid most often esteri-
fied with quinic acids such as in chlorogenic acids, is widely
found in plants [28], whereas a careful examination of litera-
ture showed that esters with shikimic acid are not commonly
found in nature despite valuable bioactivities [29].
We have recently shown that CPC is an attractive method
to isolate bioactive compounds from Solanum species [30,31].
Hence, it was decided to implement this technique to the frac-
tionation of S. somalense leaves that are a rich source of caffeic
acid derivatives with potential interest. The key points critical
for performing separation by CPC are good solubility of the
sample in the solvent system and the determination of a suit-
able two-phase system allowing a good partition behaviour of
the target compounds. The efficiency of the partition is evalu-
ated using the KD parameter. KD of the six main compounds
was determined by HPLC in several solvent systems (Table 2).
In the case of natural product studies by CPC, the sample
complexity and the unknown nature of target compounds
demand the careful consideration of the different KD values
and to depart from the usual rules of choice of solvent system.
Indeed, generally speaking, compounds of interest should be
almost equally distributed between the two phases (KD val-
ues in the range 0.5–2) and the separation factor should be
>1.5 to ensure the most successful conditions of separation
[32].
At this stage of work, our attention was focused on the
isolation of the main component of the extract (peak B,
i.e. 5-O-caffeoylshikimic acid) and the KD values were ex-
amined according to this aim. Of all the possible combina-
tions of solvents, we first chose two families of ternary sol-
vent systems, EtOAc/MeOH/H2 O and CHCl3 /MeOH/H2 O,
which accounted for >60% of the separation of glycosylated
flavonoids in the literature [33], although we knew that other
types of compounds were present in the extract. Moreover,
despite the difficulty of using BuOH solvent due to its weak
volatility, this solvent was also tested because of its ability to be
a major organic phase modifier. Based on the data collected
(Table 2), we chose to perform the experiments on the basis
of the following requirements: (i) for all the peaks (A–F), KD
values had to be in the smallest possible range to allow the
separation of all the compounds within a reasonable time,
(ii) a greater part of the values should be close to the interval
0.5–2 [34], and (iii) KD values f the six different compounds
should lead to separation factors close to 1.5 [34].
With these constraints and with the objective to tar-
get the major compound B, the system CHCl3 /MeOH/H2 O
(9.5:10:5) (KD values of 10.00, 3.34, 4.62, 4.86, 1.37, 0.16,
respectively, for peaks A–F; separation factor ␣ between com-
pound B and compound C of 1.4) seemed to be promising
for achieving the isolation of the compound B by CPC.
Crude S. somalense extract (0.82 g) was fractionated and
despite the rather uncertain nature of this initial choice, the


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2014, 37, 2331–2339 Liquid Chromatography 2337

163.03
100
A
%

337.09
0 m/z
50 100 150 200 250 300 350 400 450 500 550

335.07
100
B
%

179.04
0 m/z
50 100 150 200 250 300 350 400 450 500 550

179.04
100
C

O OH C9H7O4-
m/z 179.04
- H2O m/z 161.03
- CO2 m/z 135.05

HO O Figure 4. LC–ESI-MS spectra of com-

%

pound B and fragmentation scheme (a)

OH positive-ion mode (b) negative-ion mode,
OH and (c) MS/MS of the [M−H]− ion at m/z
335.08.Conditions: Nucleoshell RP18 col-
135.05 C9H7O3+
OH umn (100 × 2 mm, 2.7 ␮m) maintained
m/z 163.03
at 30⬚C. The elution was performed us-
161.03 ing a 0.25 mL/min mobile phase gradient
programmed from water (A), acetonitrile
335.08 (B), both containing 0.1% v/v HCOOH, as
follows (A/B): 95:5 (t = 0 min), 95:5 (t =
0 m/z 5 min), 5:95 (t = 30 min), 5:95 (t = 40 min),
100 150 200 250 300 350 95:5 (t = 45 min), and 95:5 (t = 50 min).

operation proved to be successful. Each fraction collected
during the separation process was weighed and analyzed by
HPLC (Fig. 3) showing that the fractions f19 –f24 contained
almost solely the compound B with purity varying from 86
to 97%. A total of 21.8 mg of 5-O-caffeoylshikimic acid was
obtained from the three fractions of highest purity, f21 –f23
(97% purity), leading to a yield of 2.63%. Its structure was
confirmed by NMR spectroscopy. By taking into account the
quantity of B in fractions f19 –f24 (39.3 mg), it was also pos-
sible to calculate the content of 5-O-caffeoylshikimic acid in
S. somalense leaves that is 0.74% of DW.
Studies to isolate secondary metabolites from plant mate-
rial are often unique studies and it is therefore difficult to have
reference points to compare and evaluate the performance of
natural product isolation approaches. However our results are
quite satisfactory within the context of the isolation of natural
products. The use of CPC to isolate the major compound of
S. somalense leaves has enabled us to show the great advan-
tage of this technology compared to the separation on a silica
column, which is a more conventional technique overwhelm-
ingly used for the fractionation of natural products. Indeed,
our results may be analyzed according to the two primary di-
mensions highlighted by Pauli et al. [9] in their meta-analysis
of natural products purification schemes. The first criterion
disclosed in their study is the number of purification steps.
On this point, our study helps to show the superiority of CPC
compared to silica gel column: a large amount (21.8 mg) of
5-O-caffeoylshikimic acid was obtained in only a single CPC
step, leading to a yield of 2.63% compared to 1.08% for sil-
ica gel chromatography (Fig. 3). Moreover, a major widely
known disadvantage of silica gel purification was also noted
in our study: 45% of the extract was irreversibly absorbed on
the column, while almost all the samples were recovered by
CPC. The second dimension described by Pauli et al. [9], is the


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
2338 S. Chideh et al. J. Sep. Sci. 2014, 37, 2331–2339

Table 2. KD (distribution coefficient) of six targets peaks measured in different solvent systems

Solvent system Composition A B C D E F

CHCl3 /MeOH/H2 O 04:03:02 98.37 17.00 50.03 82.98 32.43 18.87

06:07:04 39.26 10.25 18.89 18.52 13.12 10.76
08:10:05 26.20 6.08 10.95 11.03 7.15 6.55
9.5:10:5 10.00 3.34 4.62 4.86 1.37 0.16
11:10:05 8.00 2.57 2.78 2.44 2.37 1.87
13:10:05 12.75 3.44 4.23 4.23 2.86 2.21
15:10:05 41.51 6.86 11.09 11.28 7.02 5.55
12:10:05 49.73 7.19 15.97 20.29 10.25 7.80
CHCl3 /MeOH/BuOH/H2 O 42:32:5:21 22.81 3.36 5.66 5.83 3.59 2.66
42:30:8:20 13.61 2.49 3.22 3.44 2.09 1.73
45:30:5:20 24.62 3.84 5.49 5.53 3.49 2.78
40:30:10:20 11.11 2.21 2.65 2.72 1.61 1.34
37:37:4:22 17.20 4.20 5.86 5.98 3.96 3.34
38:30:12:20 10.95 2.23 2.66 2.54 1.65 1.19
EtOAc/MeOH/H2 O 40:20:40 0.04 0.51 0.25 0.21 0.31 0.38
50:00:50 0.02 0.88 0.19 0.13 0.16 0.39
47:06:47 0.03 0.83 0.19 0.09 0.23 0.42
44:11:44 0.02 0.77 0.17 0.06 0.21 0.38
40:20:40 0.04 0.54 0.30 0.24 0.34 0.48
36:18:36 0.06 0.34 0.23 0.19 0.25 0.30
EtOAc/BuOH/H2 O 50:00:50 0.03 0.53 0.14 0.08 0.18 0.33
47.5:05:47.5 0.05 0.88 0.40 0.30 0.49 0.85
45:10:45 0.12 1.42 1.47 1.28 1.95 3.01
42.5:15:42.5 0.21 1.49 2.94 2.71 3.97 5.70
AcOEt/MeOH/BuOH/H2 O 45:10:05:40 0.10 1.06 0.96 0.86 1.28 1.82

chromatographic methodology: CCC methods are used only
sporadically (average 0.9%) and despite recent developments
of this technology, its proportional use in natural product
studies has actually decreased from 1.7 to 0.3% during the
period of the survey. Through this work, we have attempted
to promote the use of CPC by providing some elements to
rationalize the solvent selection point that often restricts the
development of CPC in isolation approaches. In addition, we
showed that a KD value of 3.3 (different from 1) allowed us to
carry out the isolation of 5-O-caffeoylshikimic acid in <3 h, by
taking into account all stages from the preparation of solvents
until the recovery of the purified sample.
4 Concluding remarks
In this study, the use of CPC was highly efficient for
the production of 5-O-caffeoylshikimic acid from S. soma-
lense leaves. We were thus able to increase the yield of
5-O-caffeoylshikimic acid from 1.08 to 2.63% using the
CPC method rather than a silica gel column chromatogra-
phy. These results will enable the production of significant
amounts of this compound for further biological studies be-
cause a scaling can easily be considered with CPC. Work
on S. somalense leaves is currently being pursued by using
several approaches. The complete structural identification of
other compounds present in the extract in large quantities is
currently in progress, as well as the biological evaluation of ex-
tracts and isolated compounds in different ways, antioxidant,
microbiological, and others. In addition, taking into account
the work of Ziouti [35], the presence of 5-O-caffeoylshikimic
acid will be studied in other parts of the S. somalense plant.
All these studies will contribute to fully enhance and valorize
the medicinal resources of Djibouti.
The authors have declared no conflict of interest.
5 References
[1] Kinghorn, A. D., Pan, L., Fletcher, J. N., Chai, H., J. Nat.
Prod. 2011, 74, 1539–1555.
[2] Phillipson, J. D., Phytochemistry 2007, 68, 2960–2972.
[3] Hassan-Abdallah, A., Merito, A., Hassan, S., Aboubaker,
D., Djama, M., Asfaw, Z., Kelbessa, E., J. Ethnopharma-
col. 2013, 148, 701–713.
[4] Azmir, J., Zaidul, I. S. M., Rahman, M. M., Sharif, K. M.,
Mohamed, A., Sahena, F., Jahurul, M. H. A., Ghafoor, K.,
Norulaini, N. A. N., Omar, A. K. M., J. Food Eng. 2013,
117, 426–436.
[5] Khansari, N., Shakiba, Y., Mahmoudi, M., Recent Pat. In-
flamm. Allergy Drug Discov. 2009, 3, 73–80.
[6] Willcox, J. K., Ash, S. L., Catignani, G. L., Crit. Rev. Food
Sci. Nutr. 2004, 44, 275–95.
[7] Tu, Y., Jeffries C., Ruan, H., Nelson, C., Smithson, D.,
Shelat, A. A., Brown, K. M., Li, X-C., Hester, J. P., Smillie,


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2014, 37, 2331–2339
T., Khan, I. A., Walker, L., Guy, K., Yan, B., J. Nat. Prod.
2010, 73, 751–754.
[8] Zhang, L., Ravipati, A. S., Koyyalamudi, S. R., Jeong, S.
C., Reddy, N., Smith, P. T., Bartlett, J., Shanmugam, K.,
Meunch, G., Wu, M. J., J. Agric. Food Chem. 2011, 59,
12361–12367.
[9] Pauli, G. F., Chen, S. N., Friesen, J. B., McAlpine, J. B.,
Jaki, B. U., J. Nat. Prod. 2012, 75, 1243–1255.
[10] Dagnon, S., Ivanov, I., Bojilov, D., Docheva, M.,
Statkova, S., J. Pharmacogn. Phytochem. 2013, 1, 76–
84.
[11] Stalikas, C., J. Sep. Sci. 2007, 30, 3268–3295.
[12] Trabelsi, N., Oueslati, S., Henry-Vitra, C., Waffo-Téguo,
P., Medini, F., Mérillon, J. M., Abdelly, C., Ksouri, R., Ind.
Crop Prod. 2013, 49, 740–746.
[13] Pauli, G. F., Pro, S. M., Friesen, J. B., J. Nat. Prod. 2008,
71, 1489–1508.
[14] Berthod, A., Ruiz-Ángel, M. J., Carda-Broch, S., J. Chro-
matogr. A 2009, 1216, 4206–4217.
[15] Wang, T., Jiang, X., Yang, L., Wu, S., J. Chromatogr. A
2008, 1180, 53–58.
[16] Harborne, J. B., Williams, C. A., Greenham, J., Moyna,
P., Phytochemistry 1974, 13, 1557–1559.
[17] Zeng, X., Wang, Y., Qiu, Q., Jiang, C., Jing, Y., Qiu, G.,
He, X., Fitoterapia 2012, 83, 104–109.
[18] Wan, C., Yuan, T., Cirello, A. L., Seeram, N. P., Food Chem.
2012, 135, 1929–1937.
[19] Gan, R. Y., Xu, X. R., Song, F. L., Kuang, L., Li, H. B., J.
Med. Plants Res. 2010, 4, 2438–2444.
[20] Conway, W. D., J. Chromatogr. A 2011, 1218, 6015–
6023.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 2339
[21] Sheridan, H., Krenn, L., Jiang, R., Sutherland, I., Ig-
natova, S., Marmann, A., Liang, X., Sendker, J., J.
Ethnopharmacol. 2012, 140, 482–491.
[22] Li, H. B., Wong, C. C., Cheng, K. W., Chen, F., Food Sci.
Technol. 2008, 41, 385–390.
[23] Ali, S. S., Kasoju, N., Luthra, A., Singh, A., Sharan-
abasava, H., Sahu, A., Food Res. Int. 2008, 41, 1–15.
[24] Matkowski, A., Biotechnol. Adv. 2008, 26, 548–560.
[25] Fukuoka, M., Chem. Pharm. Bull. 1982, 30, 3219–3224.
[26] Jaiswal, R., Deshpande, S., Kunhert, N., Phytochem.
Anal. 2011, 22, 432–441.
[27] Ma, C., Dastmalchi, K., Whitaker, B. D., Kennelly, E. J., J.
Agric. Food Chem. 2011, 59, 9645–9651.
[28] Dai, J., Mumper, R. J., Molecules 2010, 15, 7313–7352.
[29] Sambanthamurthi, R., Tan, Y., Sundram, K., Abeywar-
dena, T. M., Sambandan, G., Rha, C., Sinskey, A. J., Sub-
ramaniam, K., Leow, S. S., Hayes, K. C., Wahid, M. B., Br.
J. Nutr. 2011, 106, 1655–1663.
[30] Attoumbré, J., Lesur, D., Giordanengo, P., Baltora-
Rosset, S., J. Chromatogr. B 2012, 908, 150–154.
[31] Attoumbré, J., Giordanengo, P., Baltora-Rosset, S., J.
Sep. Sci. 2013, 36, 2379–2385.
[32] Chen, L. J., Games, D. E., Jones, J., J. Chromatogr. A
2003, 988, 95–105.
[33] das Neves Costa, F., Guimarães Leitão, G., J. Sep. Sci.
2010, 33, 336–347.
[34] Severino, V. G. P., Cazal, C., de, M., Forim, M. R., da Silva,
M. F. d. G. F., Rodrigues-Filho, E., Fernandes, J. B., Vieira,
P. C., J. Chromatogr. A 2009, 1216, 4275–4281.
[35] Ziouti, A., El Modafar, C., Fleuriet, A., El Boustani, S.,
Macheix, J. J., Biol. Plant. 1996, 38, 451–457.
www.jss-journal.com