Isolation of Three Triterpene Saponins, Including Two New Oleanane Derivatives, from Soldanella alpina and Hydrophilic Interaction Liquid Chromatography–Evaporative Light Scattering Detection of these Three Saponins in Four Soldenella Species

Article abstract of DOI:10.1002/pca.2706

Introduction: The genus Soldanella is one of the few endemic to Europe. Some of its species have relevance in local traditional medicine. Earlier work has indicated the possible presence of saponins in S. alpina. Objective: To investigate S. alpina and other related species for the occurrence of saponins. Methods: Following sequential extraction with n-hexane, dichloromethane and ethyl acetate the subsequent methanolic extract of S. alpina roots was fractionated after solvent precipitation using fast centrifugal partition chromatography and column chromatography. Structures were elucidated by LC-MSⁿ, high-resolution MS, hydrolysis experiments and one-dimensional (1D)- and two-dimensional (2D)-NMR. A hydrophilic interaction liquid chromatography method was developed to quantitate saponins in the leaves and roots of four Soldanella species. Results: Three triterpene saponins, two of them new natural products, were isolated from S. alpina. Based on an epoxyoleanal aglycone substituted with four sugar units, they were analytically quantitated using a Kinetex 2.6 μm hydrophilic interaction liquid chromatography (HILIC) column together with a mobile phase comprising of ammonium acetate, water and acetonitrile. Method validation confirmed that the assay meets all requirements in respect to linearity, accuracy, sensitivity and precision. All four Soldanella species investigated contained the three saponins. The lowest total level of the three saponins (1.09%) was observed in S. montana leaves while the highest saponin content (5.14%) was determined in S. alpina roots. Conclusion: The detection of saponins within the genus Soldanella is an indication that further phytochemical examination of this genus may reveal more secondary metabolites of interest. Copyright

Full text of DOI:10.1002/pca.2706

Research Article
Received: 15 December 2016,
Revised: 24 May 2017,
Accepted: 25 May 2017
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI 10.1002/pca.2706
Isolation of Three Triterpene Saponins,
Including Two New Oleanane Derivatives, from
Soldanella alpina and Hydrophilic Interaction
Liquid Chromatography–Evaporative Light
Scattering Detection of these Three Saponins in
Four Soldenella Species
Julia Haller,a† Stefan Schwaiger,a† Hermann Stuppner,a Frank Gafnerb
and Markus Ganzeraa*
ABSTRACT:
Introduction – The genus Soldanella is one of the few endemic to Europe. Some of its species have relevance in local traditional
medicine. Earlier work has indicated the possible presence of saponins in S. alpina.
Objective – To investigate S. alpina and other related species for the occurrence of saponins.
Methods – Following sequential extraction with n-hexane, dichloromethane and ethyl acetate the subsequent methanolic extract
of S. alpina roots was fractionated after solvent precipitation using fast centrifugal partition chromatography and column
chromatography. Structures were elucidated by LC-MSn, high-resolution MS, hydrolysis experiments and one-dimensional
(1D)- and two-dimensional (2D)-NMR. A hydrophilic interaction liquid chromatography method was developed to quantitate
saponins in the leaves and roots of four Soldanella species.
Results – Three triterpene saponins, two of them new natural products, were isolated from S. alpina. Based on an epoxyoleanal
aglycone substituted with four sugar units, they were analytically quantitated using a Kinetex 2.6 μm hydrophilic interaction
liquid chromatography (HILIC) column together with a mobile phase comprising of ammonium acetate, water and acetonitrile.
Method validation confirmed that the assay meets all requirements in respect to linearity, accuracy, sensitivity and precision.
All four Soldanella species investigated contained the three saponins. The lowest total level of the three saponins (1.09%) was
observed in S. montana leaves while the highest saponin content (5.14%) was determined in S. alpina roots.
Conclusion – The detection of saponins within the genus Soldanella is an indication that further phytochemical examination of
this genus may reveal more secondary metabolites of interest. Copyright © 2017 John Wiley & Sons, Ltd.
Additional Supporting Information may be found online in the supporting information tab for this article.
Keywords: Soldanella; triterpene saponins; hydrophilic interaction liquid chromatography; isolation; quantification
has been filed describing a cosmetic preparation against skin
ageing, which contains a S. alpina extract as the active ingredient
Introduction
The genus Soldanella L. (Primulaceae) is one of only 27 genera
endemic to Europe and it comprises 16 species found in the
Alps and mountainous parts of southern Europe (e.g. Pyrenees,
Apennine, Balkan). The rather small (3–20 cm) perennial plants
are known as snowbells (English) or Alpenglöckchen (German).
They have a basal rosette of simple, orbicular leaves, with
flower stalks arising from the centre of the rosette, each stalk
bearing one to six white to violet flowers (Zhang and Kadereit,
2002). One of the more widely distributed species is Soldanella
alpina, a plant sometimes also cultivated for ornamental
purposes (it was the flower of the year 2004 in Germany).
Soldanella alpina has been used as a model organism to study
the effects of high temperature or light stress on photoinhibition
(Streb et al., 2003a; Laureau et al., 2015), on antioxidative
scavenging capacity (Laureau et al., 2011) and the occurrence of
metabolites like ascorbate or malate (Streb et al., 2003b). A patent
(Dudler and Stangl, 2015). The plant is traditionally used as a
sedative drug in some parts of Switzerland (Süßmuth, 2013).
However, it should be noted that the term Herba Soldanellae
was also used for Convolvulvus soldanella (Kosch, 1939) and
Brassica marina (Frerichs et al., 1949) in former times.
* Correspondence to: Markus Ganzera, Institute of Pharmacy, Pharmacognosy,
University of Innsbruck, Innrain 80-82, 6020 Innsbruck, Austria. Email:
markus.ganzera@uibk.ac.at
†Julia Haller and Stefan Schwaiger contributed equally.
a
Institute of Pharmacy, Pharmacognosy, Member of the CMBI, University of
Innsbruck, Innrain 80-82, 6020, Innsbruck, Austria
b
Mibelle Biochemistry, Bolimattstrasse 1, 5033, Buchs, Switzerland
Phytochem. Anal. 2017
Copyright © 2017 John Wiley & Sons, Ltd.

Haller J. et al.
Previous investigations described the composition of epicuticular
waxes of S. pusilla (Lütz and Gülz, 1985) and the presence of
flavonoids in S. alpina (Kroslakova et al., 2016). The occurrence of
saponins in S. alpina has already been suspected more than 90
years ago (Luft, 1926). Since S. alpina has certain relevance in
traditional medicine, this present study investigated the
occurrence of saponins in this and three other Soldanella species.
HR-ESI-MS of isolated compounds
Experiments were performed on a micrOTOF-QII mass spectrometer
(Bruker Daltronics, Bremen, Germany) by directly infusing a methanolic
solution of each compound (0.25 mg/mL). MS conditions were set to
negative electrospray ionisation, a nebuliser pressure of 4.4 psi, a HV
capillary voltage of 3.5 kV, 180°C dry temperature, a dry gas flow of 4
L/min and a scan range from m/z 100 to 3000, respectively.
Sample preparation
Experimental
For preparation of the HPLC sample solutions the finely powdered plant
material (root or leaf, 50 mg) was extracted with methanol (3 mL) by
sonication (15 min). The mixture was centrifuged (5 min, 3500 rpm) and
the supernatant placed in a 10 mL volumetric flask. This extraction step
was repeated two more times, solutions were combined and the flask filled
to volume with methanol. Sample solutions are stable for at least two
weeks if stored at 4°C. Prior to HPLC analysis an aliquot of the leaf extract
(5.00 mL) or the root extract (10.00 mL) was evaporated to dryness under
reduced pressure, the residue was re-dissolved in methanol (1.00 mL) and
membrane filtered (GHP Acrodisc 13, 0.45 μm polypropylene membrane,
Pall, Port Washington, NY, USA).
Plant material and chemicals
Three individual samples (i.e. 5–10 entire plants each) of Soldanella alpina
(SA-1 to SA-3) as well as single samples of S. pusilla (SP-1), S. x transsylvanica
(ST-1) and S. montana (SM-1) were analysed. All of them, except SA-3
(Mediplant, Conthey, Switzerland) came from Gärtnerei Eschmann, Emmen,
Switzerland. They were harvested in May 2010 (SA-1), June 2011 (SA-2 and
SA-3) and May 2012 (SP-1, ST-1 and SM-1). The plant material was
authenticated by one of the authors (S. Schwaiger), separated into roots
(R) and leaves (L), and dried at ambient temperature. Voucher specimens
of all samples are deposited at the Institute of Pharmacy, Pharmacognosy,
University of Innsbruck.
All chemicals required for isolation and analysis were of analytical grade
(p.a.) quality and were purchased from Merck (Darmstadt, Germany).
Deuterated pyridine for NMR experiments came from Euroiso-top (Saint-
Aubin Cedex, France). HPLC grade water (18.0 MΩ/cm) was produced by
a Satorius Arium 611 water purification system (Göttingen, Germany).
Analytical conditions
HPLC experiments were performed on an Agilent 1200 system (Agilent,
Waldbronn, Germany) equipped with binary pump, autosampler and
column heater, fitted with a Sedex 85 LT ELSD detector from Sedere
(Alfortville Cedex, France). The best separation was achieved using a
Kinetex HILIC 2.6 μm 100 Å column (150 mm × 4.6 mm, Phenomenex,
Torrance, CA, USA), utilising a mobile phase comprising 20 mM ammonium
acetate in water (A), and a 9:1 mixture (v/v) of acetonitrile and 200 mM
ammonium acetate solution (B). Both solutions were adjusted to a pH 4.0
with acetic acid. The linear gradient started at 3% A/97% B to end after
25 min at 8% A/92% B. The column was washed for 5 min with 30%
A/70% B before re-equilibration for 20 min. Column temperature, flow rate
and sample volume were set to 30°C, 0.8 mL/min and 5 μL, respectively. The
evaporative light scattering detection (ELSD) settings were 40°C, a nebuliser
pressure of 3.6 bar (nitrogen) and gain 12.
Isolation of triterpene saponins from S. alpina
Sequential extraction (5 × 100 mL, 10 min ultrasonic bath for each solvent)
of dried and powdered S. alpina roots (8.5 g, sample SA-3-R) and
subsequent evaporation yielded n-hexane (45.8 mg), dichloromethane
(39.7 mg), ethyl acetate (18.6 mg) and methanol (2304.4 mg) extracts. The
methanolic extract (ca. 2 g) was dissolved in methanol/water (1:1; v/v, 4.0
mL) and mixed with acetone (8.0 mL). The resultant precipitate was
separated by decantation to yield fraction 1 (17 mg). The remaining
solution was mixed with more acetone (4 mL) and kept at room
temperature for 24 h. During that time the solution separated into an oily
syrup (at the bottom of the beaker) and a light yellow solution that were
separated and evaporated to dryness to give fraction 2 (904.0 mg) and
fraction 3 (605.6 mg), respectively. Fraction 3 was subsequently dissolved
in methanol (8.0 mL) and mixed with diethyl ether (8.0 mL) resulting in a
white precipitate (246.5 mg, fraction 3p) and the remaining supernatant,
which yielded fraction 3s (383.6 mg) after solvent removal. Fraction 3s
was further purified by centrifugal partition chromatography (FCPC A200,
Method validation
A standard stock solution of the three saponins (5 mg each) was prepared
with methanol (5.00 mL). Further dilutions were prepared in the ratio of 1:2
with the same solvent. The limits of detection (LOD) and limits of
quantitation (LOQ) were visually evaluated, defining concentrations with
a peak height of 3- or 10-times the baseline noise, respectively. Accuracy
was investigated by spiking sample SA-2-L with different concentrations
of the standards (high, medium and low spike). The quantitative results
were compared with the theoretically present amount and expressed as
recovery rate. Precision was investigated by analysing five individually
prepared solutions of sample SP-1-L on day 1. On days 2 and 3 the same
procedure was repeated, and the variation within one day (intra-day
precision) and within three days (inter-day precision) was calculated based
on peak area.
Kromaton, Angeres, France) using
a
solvent system of
chloroform/methanol/water/1-propanol (9:12:8:1; v/v/v/v). The lower phase
was pumped through the system (0.5 mL/min) as mobile phase in
descending mode with counter clockwise rotation of 800 rpm. Fractions
(2 mL) were collected and checked by HPTLC (LiChrospher silica gel 60
F254 plates, 7 μm, Merck, Darmstadt, Germany) using a developing solution
of chloroform/methanol/water/formic acid (6:3.2:0.8:1.2; v/v/v/v) and
visualised by spraying and heating with anisaldehyde-sulphuric acid
reagent. The pooled fractions (fraction 3sA) collected between 160 to 288
mL contained a mixture of two saponins, while the fraction eluting between
362 and 382 mL contained a pure saponin (13.8 mg, Sap3; hRf 50.0).
An aliquot (78.6 mg) of fraction 3sA was further separated by silica gel
column chromatography (diameter: 2.5 cm, length: 30 cm) using an
isocratic elution with chloroform/methanol (1:1, v/v). Fractions (3 mL) were
monitored using HPTLC yielding Sap1 (eluting between 180 and 225 mL,
36.3 mg, hRf 54.4) and Sap2 (eluting between 270 and 360 mL, 17.4 mg,
hRf 52.2). Purity and identity of the isolated compounds were verified by
HPTLC, LC-MS3, GC-FID, NMR and HR-ESI-MS (see Supporting Information
for LC-MSn and one-dimensional (1D)- and two-dimensional (2D)-NMR
spectra as well as details on hydrolysis/GC-FID experiments).
Results and discussion
Structural elucidation of saponins in S. alpina
Firstly, the three isolated saponins (Fig. 1) were studied by LC-ESI-
MS/MS. The parent ions ([M-H]˗) were observed at m/z 1059.9
(Sap1 and Sap2) and 1075.9 (Sap3). Taking Sap1 as an example,
further fragmentation showed signals at m/z 927.6 (MS2) and
765.1, 603.1 and 471.9 (MS3). The fragmentation pattern suggested
the loss of four sugar units as displayed by observed mass-
wileyonlinelibrary.com/journal/pca
Copyright © 2017 John Wiley & Sons, Ltd.
Phytochem. Anal. 2017

Isolation and analysis of saponins in Soldanella species
HMBC correlations, while the α- or β-orientation was deduced from
the corresponding coupling constants of anomeric protons (J <
5.1 Hz, α-orientation; J > 6.8 Hz β-orientation). Respective NMR
data showed similar chemical shift values except for the terminal
aldopentose in Sap2. Common features were an initial α-L-
arabinose unit which is linked to two β-D-glucose units. The
terminal pentose was α-L-arabinose in Sap1 and Sap3 while for
Sap2 it was β-D-xylose. The NMR data of all isolated compounds
are compiled in Tables 1 (1H-NMR) and Tables 2 (13C-NMR).
In order to verify the composition of the individual sugar chains,
hydrolysis and GC-experiments were performed. Each saponin was
subjected to acidic hydrolysis, the reaction product was
derivatised and analysed by GC in comparison to standard
compounds (see Supporting Information for experimental details).
For Sap1 and Sap3 only signals corresponding to L-(+)-arabinose
and D-(+)-glucose were found, while Sap2 revealed the presence
of L-(+)-arabinose, D-(+)-xylose and D-(+)-glucose (Figs S4–S6).
Accordingly, Sap1 was elucidated as 3β,16α-dihydroxy-
13β,28-epoxyolean-30-al 3-O-[(O-β-D-glucopyranosyl-(1→2)-
O-[O-α-L-arabinopyranosyl-(1→2)-β-D-glucopyranosyl-(1→4)]-
α-L-arabinopyranoside)], Sap2 as 3β,16α-dihydroxy-13β,28-
Figure 1. Chemical structure of the main triterpene saponins isolated from
Soldanella alpina.
differences of 132 and 162. This indicated that the structure of
Sap1 contains two hexoses and two pentoses. Sap2 and Sap3
were analysed accordingly (see Supporting Information, Figs S1–
S3). Since the mass of the aglycon of Sap3 differed by 16 to those
of Sap1 and Sap2 the presence of an additional oxygen atom in
the structure was assumed.
epoxyolean-30-al
3-O-[(O-β-D-glucopyranosyl-(1→2)-O-
[O-β-D-xylopyranosyl-(1→2)-β-D-glucopyranosyl-(1→4)]-α-
L-arabinopyranoside)], and Sap3 as 3β,16α,23-trihydroxy-
13β,28-epoxyolean-30-al 3-O-[(O-β-D-glucopyranosyl-(1→2)-O-
[O-α-L-arabinopyranosyl-(1→2)-β-D-glucopyranosyl (1→4)]-α-L-
arabinopyranoside)]. The deduced molecular formulae were
confirmed by HR-ESI-MS revealing for Sap1 and Sap2 an m/z
value of 1059.5413 (calculated [M-H]˗ 1059.5381, mass error 3.2
ppm) and for Sap3 m/z 1075.5366 (calculated [M-H]˗ 1075.5331,
mass error 3.5 ppm). Sap2 has already been isolated as
deglucocyclamin from Cyclamen hederifolium (Altunkeyik et al.,
2012) and as ardisiacrispin A from several sources including
Ardisia crispa ( Jansakul et al., 1987), Lysimachia foenum-graecum
(Dai et al., 2017) and Labisia pumila (Avula et al., 2011). The latter
publication also reports on a quantitative HPLC-ELSD method.
Sap1 and Sap3 have not previously been described.
The 1D- (1H, 13C) and 2D-NMR spectra (COSY, HSQC, HMBC,
NOESY) of the saponins revealed for all three compounds a similar
oleanan-type aglycon with an aldehyde function at position 30, a
hydroxyl-group at position 16 and an additional ring formation
(position 13 to 17) via a -O-CH2- unit. While Sap1 and Sap2
showed no further structural differences, Sap3 contained an
additional hydroxyl-group at position 23, which was differentiated
from position 24 by long-range 1H-1H-COSY and NOESY
correlations between position 24 and 25. Connectivity and
structure of the sugar units were established by 1H-1H-COSY and
Table 1. 1H-NMR data of isolated compounds (in pyridine-d5; 600.19 MHz; σ in ppm; J in parentheses)
Position
1
Sap1
Sap2
Sap3
Ha 1.66 d (12.2)
Hb 0.89 d (12.5)
Ha 1.83 dd (12.8, 25.4)
Hb 2.02 d (12.5)
3.18a, 1H
Ha 1.67 d (11.4)
Hb 0.89 d (13.0)
Ha 2.03 br d (12.6)
Hb 1.85 q (12.5)
3.19a, 1H
—
Ha 1.73a
Hb 1.00a
2
Ha 2.12a
Hb 1.97a
3
4
5
6
4.11a, 1H
—
—
0.70 d (11.6), 1H
0.70 d (11.4), 1H
1.54 d (12.7), 1H
Ha 1.44a
Ha 1.45a
Ha 1.69a
Hb 1.40 d (15.0)
Ha 1.55a
Hb 1.40a
Hb 1.45a
7
Ha 1.50a
Ha 1.69a
Hb 1.22a
Hb 1.22a
Hb 1.20 d (12.2)
—
8
9
10
11
—
—
1.28 d (12.2), 1H
—
1.28a, 1H
—
1.40a, 1H
—
Ha 1.75 dq (4.5, 13.6, 13.9)
Ha 1.75 m
Ha 1.78 dd (2.8, 12.7)
Hb 1.44a
Hb 1.45a
Hb 1.46a
12
Ha 2.12a
Ha 2.11a
Ha 2.09a
(Continues)
Phytochem. Anal. 2017
Copyright © 2017 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/pca

Haller J. et al.
Table 1. (Continued)
Posit on
Position
Sap1
Sap2
Sap3
Hb 1.44a
—
Hb 1.44a
—
Hb 1.44a
—
13
14
15
—
—
—
Ha 2.21 dd (4.5, 14.6)
Hb 1.50 d (14.9)
4.22a, 1H
Ha 2.21 dd (3.7, 14.1)
Ha 2.19 dd (4.3, 14.0)
Hb 1.51a
Hb 1.42a
16
17
18
19
4.23a, 1H
4.19a, 1H
—
—
—
1.40 d (15.0), 1H
Ha 2.87 t (13.4)
Hb 2.13 m
1.41a, 1H
1.39a, 1H
Ha 2.87 t (13.3)
Ha 2.84 t (13.4)
Hb 2.14a
Hb 2.10a
20
21
—
—
—
Ha 2.57 dt (4.2, 13.1, 13.5)
Hb 2.10 m
Ha 1.98 dd (3.0, 13.9)
Hb 1.60 dd (4.4, 13.5)
1.24 s, 3H
Ha 2.57 dd (4.7, 13.5)
Hb 2.11 m
Ha 1.98 dd (4.2, 13.3)
Hb 1.60 dd (4.7, 13.6)
1.24 s, 3H
Ha 2.54 dt (3.7, 12.7, 13.1)
Hb 2.08a
22
23
Ha 1.95a
Hb 1.60 dd (5.2, 13.7)
Ha 4.28a
Hb 3.74a
24
25
26
27
28
1.10 s, 3H
0.85 s, 3H
1.31 s, 3H
1.56 s, 3H
Ha 3.56 d (7.2)
Hb 3.19 m
1.04 s, 3H
9.66 s, 1H
1.11 s, 3H
0.86 s, 3H
1.31 s, 3H
1.57 s, 3H
Ha 3.56 d (6.9)
Hb 3.18 m
1.04 s, 3H
9.65 s, 1H
1.09 s, 3H
0.95 s, 3H
1.33 s, 3H
1.49 s, 3H
Ha 3.55 d (7.4)
Hb 3.17 d (7.4)
1.00 s, 3H
9.63 s, 1H
29
30
α-L-Arabinose unit at position 3
1′
2′
3′
4′
4.81 d (4.5), 1H
4.57a, 1H
4.82 d (5.1), 1H
4.63a, 1H
5.04 d (4.5)a, 1H
4.61a, 1H
4.30a, 1H
4.30a, 1H
4.21a, 1H
4.27a, 1H
4.30a, 1H
Ha 4.58 d (12.9)
Hb 3.69 d (11.7)
4.22a, 1H
5′
Ha 4.64 dd (3.1, 7.3)
Hb 3.69 d (11.7)
Ha 4.58a
Hb 3.60 d (11.9)
β-D-Glucose unit at position 2′
1″
2″
3″
4″
5″
6″
5.50 d (6.8), 1H
4.10 t (7.5), 1H
4.27a, 1H
5.52 d (7.2), 1H
4.09 t (7.6), 1H
4.25 d (9.9), 1H
4.22a, 1H
5.51 d (7.6), 1H
4.13a, 1H
4.27a, 1H
4.26a, 1H
4.05a, 1H
Ha 4.58a
4.24a, 1H
4.04a, 1H
4.06a, 1H
Ha 4.57a
Ha 4.56a
Hb 4.43a
Hb 4.44a
Hb 4.44a
β-D-Glucose unit at position 4′
1″′
2″′
3″′
4″′
5″′
6″′
5.02 d (7.4), 1H
3.94 t (6.9), 1H
4.22a, 1H
5.08 d (7.9), 1H
4.01 dd (17.6, 24.7), 1H
4.22a, 1H
5.04 d (8.2)a, 1H
3.99 m, 1H
4.23a, 1H
4.22a, 1H
4.22a, 1H
4.24a, 1H
3.81 t (7.0), 1H
3.80 br s, 1H
3.79 m, 1H
Ha 4.45a;
Ha 4.44a
Ha 4.40a
Hb 4.32a
Hb 4.32a
Hb 4.32a
Terminal pentose unit at position 2″′
1″″
2″″
3″″
4″″
5″″
4.93 d (3.2), 1H
5.10 d (7.9), 1H
4.54a, 1H
5.00 br sa, 1H
4.05a, 1H
4.17a, 1H
4.23a, 1H
Ha 4.59a
4.04a, 1H
4.15 dd (8.9, 12.7), 1H
4.22a, 1H
4.16 dd (17.6, 23.0), 1H
4.32a, 1H
Ha 4.57a
Ha 4.66 d (12.4)
Hb 3.89 d (12.0)
Hb 3.73 t (10.9)
Hb 3.73a
aSignals overlapping.
wileyonlinelibrary.com/journal/pca
Copyright © 2017 John Wiley & Sons, Ltd.
Phytochem. Anal. 2017

Isolation and analysis of saponins in Soldanella species
Table 2. 13C-NMR data of isolated compounds (in pyridine-d5;
150.91 MHz; in ppm)
Hydrophilic interaction liquid chromatography (HILIC)–
evaporative light scattering detection (ELSD) analysis
Since the saponins possessed basically no UV absorption, even
close to 200 nm, ELSD was investigated under reversed phase
HPLC conditions using C-8 (Zorbax Eclipse XDB-C8, 5 μm particle
size), C-12 (Synergi Max-RP, 4 μm), C-18 (Synergi Fusion-RP, 4
μm) or pentafluorophenyl (Kinetex PFP, 2.6 μm) stationary phases.
However, on all these materials the compounds co-eluted. With an
amino column (Supelcosil LC-NH2, 3 μm) operated in normal phase
mode at least some peak splitting was noticed, yet the baseline
separation of all three saponins was only possible on hydrophilic
interaction liquid chromatography (HILIC) material (Fig. 2). After
modifying HPLC parameters (see later) the required analysis time
was less than 13 min and signals with excellent peak shape and
resolution were obtained.
HILIC phases are polar and most commonly based on silica
material or they show zwitterionic characteristics (Zuo et al.,
2014). Both modes were tested and an unbonded silica core-shell
material (Kinetex HILIC 2.6 μm 100 Å) yielded better results with
respect to resolution and repeatability than the zwitterionic mode
(ZIC-HILIC 3.5 μm 100 Å column). Since HILIC is mainly based on
liquid-liquid chromatography, columns generally require longer
equilibration periods (e.g. 20 min in the current case) compared
to reversed phase (RP)-HPLC. A mobile phase typical for HILIC
containing acetonitrile and aqueous ammonium acetate was best
suited. However, pH and buffer molarity needed to be finely tuned
to achieve the desired separation, because both parameters have a
significant influence on the activity of the silanol groups of the
stationary phase. Adjusting the mobile phase to pH 5.0 resulted
in prolonged analysis time, at pH 3.0 the saponin signals partially
overlapped; thus a pH of 4.0 was selected. A higher buffer molarity
resulted in a better separation but also broader signals, by utilising
a solvent gradient peak symmetry could be improved again. The
influence of column temperature was as expected (i.e. prolonged
retention at lower temperatures); however, an unexpected
observation was made when varying the ELSD nebuliser
temperature. At 40°C peak height and symmetry were much more
enhanced than at 70°C. The saponins are obviously thermo-labile
to a certain extent, so that the unique low temperature technique
of the utilised ELSD was favourable to facilitate a more sensitive
detection. Selected chromatograms indicating the impact of
Position
Sap1
Sap2
Sap3
1
2
3
4
5
6
7
8
39.65 t
27.05 t
89.52 d
40.20 s
56.17 s
18.41 t
34.82 t
43.01 s
50.89 d
37.28 s
19.62 t
34.14 t
86.82 s
45.07 s
37.35 t
77.37 d
44.50 s
53.80 d
33.85 t
48.79 s
30.96 t
32.82 t
28.56 q
17.12 q
16.86 q
19.00 q
20.25 q
78.28 t
24.60 q
208.03 d
39.65 t
27.05 t
89.50 d
40.18 s
56.15 s
18.38 t
34.79 t
42.98 s
50.87 d
37.26 s
19.60 t
33.12 t
86.79 s
45.07 s
37.33 t
77.35 d
44.48 s
53.78 d
33.83 t
48.77 s
30.93 t
32.80 t
28.53 q
17.07 q
16.85 q
18.98 q
20.22 q
78.11 t
24.57 q
207.99 d
39.62 t
26.44 t
82.86 d
44.13 s
48.33 s
18.11 t
34.48 t
43.00 s
50.96 d
37.26 s
19.64 t
33.14 t
86.79 s
45.04 s
37.30 t
77.33 d
44.45 s
53.80 d
33.79 t
48.74 s
30.89 t
32.77 t
65.41 t
13.71 q
17.41 q
18.99 q
20.15 q
78.08 t
24.49 q
207.96 d
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
α-L-Arabinose unit at position 3
1′
2′
3′
4′
5′
105.15 d
80.22 d
73.74 d
79.08 d
64.68 t
105.19 d
80.55 d
73.78 d
78.39 d
64.82 t
104.11 d
80.75 d
73.81 d
78.56 d
64.54 t
β-D-Glucose unit at position 2′
1″
2″
3″
4″
5″
6″
105.38 d
76.72 d
78.67 d
72.31 d
76.59 d
63.48 t
105.42 d
76.83 d
78.60 d
72.34 d
78.63 d
63.82 t
105.60 d
76.69 d
78.76 d
71.88 d
76.61 d
63.14 t
β-D-Glucose unit at position 4′
1″′
2″′
3″′
4″′
5″′
6″′
104.67 d
85.89 d
78.05 d
71.57 d
78.83 d
62.80 t
104.51 d
84.63 d
78.60 d
71.67 d
78.81 d
62.82 t
104.49 d
85.89 d
78.28 d
71.52 d
78.82 d
62.75 t
Terminal pentose unit at position 2″′
Figure 2. Separation of the three standard compounds Sap1 (1), Sap2 (2)
and Sap3 (3) under optimised conditions (column: Kinetex HILIC 2.6 μm
100 Å, 150 mm × 4.6 mm; mobile phase: 20 mM ammonium acetate in
water with pH 4.0 (A), 9:1 mixture of acetonitrile and 200 mM ammonium
acetate with pH 4.0 (B); gradient: 3A/97B in 25 min to 8A/92B; temperature:
30°C; flow rate. 0.8 mL/min; sample volume: 5 μL; evaporative light
scattering detection (ELSD): 40°C, 3.6 bar (nitrogen), gain 12).
1″″
2″″
3″″
4″″
5″″
108.15 d
78.45 d
71.18 d
78.13 d
67.95 t
107.37 d
74.02 d
74.59 d
69.28 d
67.44 t
108.14 d
78.28 d
71.17 d
78.10 d
67.96 t
Phytochem. Anal. 2017
Copyright © 2017 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/pca

Haller J. et al.
individual separation parameters are shown in the Supporting
Information.
Stuppner, 2005). Thus, the logarithmic values for A and m will show
a linear trend (log A = a + xlog m). Considering these facts resulted
in correlation coefficients for all three saponins higher than 0.998,
within a linear range from at least 1000 to 37 μg/mL. The
determined LOD (≤ 6.7 μg/mL) and LOQ (≤ 20.3 μg/mL) values
are higher than those usually achievable by UV-vis (if compounds
show absorbance) or MS, but they are typical for this detection
technique. Selectivity of the method was assured by two facts.
First, all relevant signals in the samples were symmetrical without
any shoulders. Second, as an ELSD does not provide any
Method validation
The developed HILIC method was validated for linearity, LOD and
LOQ, selectivity, accuracy and precision (Table 3). For ELSD the
observed peak area (A) is related to the quantity of analyte on-
column (m) through the relationship A = amx, where (x) is the slope
of the response line and (a) is a response factor (Ganzera and
Table 3. Validation data of the developed hydrophilic interaction liquid chromatography (HILIC) method.
Parameter/compound
Sap1
Sap2
Sap3
Regression equationa
Correlation coefficient
Rangeb
y = 1.562x – 1.543
0.9994
1050–12.9 (5250–64.5)
y = 1.508x – 1.396
0.9986
1080 – 13.3 (5400 - 66.5)
y = 1.483x – 1.550
0.9991
1000 – 37.0 (5000 - 185.0)
Limit of detection (LOD)b
Limit of quantification (LOQ)b
Accuracy (high spike)c
Accuracy (medium spike)c
Accuracy (low spike)c
Precision (intra-day)d
Precision (inter-day)e
3.9 (19.5)
12.9 (64.5)
103.5
98.1
97.3
4.0 (20.0)
13.3 (66.5)
102.0
98.7
97.4
6.7 (33.5)
20.3 (101.5)
100.9
99.2
100.4
6.38
2.83
5.73
4.16
5.38
3.82
aFor evaporative light scattering detection (ELSD): y = log (peak area), x = log (concentration in μg/mL).
bIn μg/mL (absolute amounts in ng based on 5 μL injection volume in parenthesis).
cExpressed as recovery rates in percentage.
dMaximum deviation within one day based on peak area in percentage (n = 4).
eDeviation within three days based on peak area in percentage.
Figure 3. Comparison of root (R) and leaf (L) extracts of Soldanella alpina (sample SA-1), S. pusilla (SP-1), S. x transsylvatica (ST-1) and S. montana (SM-1)
analysed by hydrophilic interaction liquid chromatography (HILIC)–evaporative light scattering detection (ELSD); separation conditions as for Fig. 2.
wileyonlinelibrary.com/journal/pca
Copyright © 2017 John Wiley & Sons, Ltd.
Phytochem. Anal. 2017

Isolation and analysis of saponins in Soldanella species
spectroscopic data to show peak purity based on an UV-spectra,
LC-MS experiments were performed (see Supporting Information
for conditions). The HILIC method comprised only volatile
chemicals, thus it could readily be hyphenated with an iontrap
MS without any modifications. Respective results showed no
indications for co-eluting compounds (data not shown in detail).
Even if detection of the saponins would have been possible with
MS as well, in this study we selected the more economic option
(ELSD) because it might easier be available in many laboratories
and it equally was suitable for the analysis of saponins in Soldanella
samples.
For determination of the methods accuracy recovery
experiments were performed. Individually weighed portions of
sample SA-2-L were spiked with three concentrations of the
saponins (high spike: 120 μg/mL per compound, medium spike:
80 μg/mL, low spike: 40 μg/mL). After extraction and analysis,
maximum deviations between theoretical and observed values
of 97.3% (Sap1, low spike) and 103.5% (Sap1, high spike) were
found. Finally, intermediate precision was assured by analysing
individually prepared solutions of SP-1-L over a three-day period.
Within one day the results varied by no more than 6.4% (Sap1, n
= 5), inter-day precision was even below 4.2% (Sap2). Accordingly,
all of the required validation criteria were fulfilled, so that the
analytical procedure could be used for a reliable quantification of
saponins in diverse Soldanella samples.
Figure 4. Content of saponins (in percentage by weight) in Soldanella
alpina (SA), S. pusilla (SP), S. x transsylvanica (ST) and S. montana (SM)
samples: (A) leaves and (B) roots. Relative standard deviations (n = 3) are
indicated with bars.
Quantification of samples
Three snowbell species and one hybrid, including three samples of
S. alpina, were available for quantitative studies. To test the
extraction efficiency, two samples (SA-1-R and SA-1-L) were
treated as described under “sample preparation” and then re-
extracted once more with methanol. Comparison of the two
sequential extracts showed no quantifiable amounts of the three
target analytes in these extracts.
The HILIC–ELSD chromatograms of the methanolic extracts of S.
alpina, S. pusilla, S. x transsylvanica and S. montana (Fig. 3) were
quantitated against the calibration curves (Fig. 4). Each sample
solution was assayed in triplicate and a maximum relative standard
deviation of 4.1% (Sap3 in SM-1-L) was observed. All samples
investigated, regardless if roots or leaves, contained the three
saponins, with Sap1 always being the dominant compound (from
0.58% in SA-2-R to 2.52% in SA-3-R). In all S. alpina specimens Sap2
(up to 0.67% in leaves and 1.81% in roots) was the second most
abundant saponin. However, in the other species, particularly in
root material, Sap3 (up to 0.47% in S. pusilla roots) was taking this
position. In most samples the total saponin content was higher in
leaves compared to roots (e.g. in SA-2: 2.43% vs. 1.44%, in ST-1:
2.76% vs. 1.60%), except in two S. alpina samples which contained
a higher percentage of saponins in the roots. In this respect it
should be noted that the major compounds eluting right after
Acknowledgements
The authors thank Jakob Eschmann (Gärtnerei Eschmann, Emmen,
Switzerland) and Christoph Carlen (Mediplant Swiss Research
Centre on Medicinal and Aromatic Plants, Conthey, Switzerland)
for providing some of the plant samples, and Sören Heiderstädt
(University of Innsbruck) for his technical assistance during the
isolation of the saponins.
Declaration of interest statement
The authors have declared no conflict of interest.
References
Altunkeyik H, Gülcemal D, Masullo M, Alankus-Caliskan O, Piacente S,
Karayildirim T. 2012. Triterpene saponins from Cyclamen hederifolium.
Phytochemistry 73: 127–133.
Avula B, Wang YH, Ali Z, Smilie TJ, Khan IA. 2011. Quantitative determination
of triterpene saponins and alkenated-phenolics from Labisia pumila
using an LC-UV/ELSD method and confirmation by LC-ESI-TOF. Planta
Med 77: 1742–1748.
Dai LM, Huang RZ, Zhang B, Hua J, Wang HS, Liang D. 2017. Cytotoxic
triterpenoid saponins from Lysimachia foenum-graecum. Phytochemistry
136: 165–174.
Dudler B, Stangl D. 2015. Preparation to protect from extrinsic and intrinsic
skin aging. Patent, Germany, DE 102014216022A1.
Frerichs G, Arends G, Zörnig H. 1949. Hagers Handbuch der
Pharmazeutischen Praxis – 2nd Volume. Springer: Berlin; 752.
Ganzera M, Stuppner H. 2005. Evaporative light scattering detection for the
analysis of natural products. Curr Pharm Anal 1: 135–144.
Jansakul C, Baumann H, Kenne L, Samuelsson G. 1987. Ardisiacrispin A and
B, two utero-contracting saponins from Ardisia crispa. Planta Med 53:
405–409.
Sap3 are saccharides. Saccharose (Rt
= 14.9 min) was
unambiguously identified by NMR after isolation but the amount
of the second substance (Rt = 16.5 min) was not sufficient for an
exact structural determination. Preliminary NMR and MS-data
indicate the trisaccharide kestose. As it was not relevant for this
study, neither the differentiation of 1- or 6-kestose nor the
quantification of these two sugars was considered. The highest
overall saponin content (5.14%) was found in one of the S. alpina
samples. However, the content seems rather variable in this
species as another root sample also harvested in June 2011 only
contained 1.44% total saponins.
Kosch A. 1939. Handbuch der deutschen Arzneipflanzen. Springer: Berlin; 276.
Phytochem. Anal. 2017
Copyright © 2017 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/pca

Haller J. et al.
Kroslakova I, Pedrussio S, Wolfram E. 2016. Direct coupling of HPTLC with
MALDI-TOF MS for qualitative detection of flavonoids on phytochemical
fingerprints. Phytochem Anal 27: 222–228.
Laureau C, Bligny R, Streb P. 2011. The significance of glutathione for
photoprotection at contrasting temperatures in the alpine plant species
Soldanella alpina and Ranunculus glacialis. Physiol Plantarum 143:
246–260.
antioxidant levels in the two high mountain plant species Soldanella
alpina and Ranunculus glacialis. J Exp Bot 54: 405–418.
Süßmuth A. 2013. Lexikon der Alpenheilpflanzen, 2nd edn. AT Verlag: Aarau;
101–103.
Zhang LB, Kadereit JW. 2002. The systematics of Soldanella L. (Primulaceae)
based on morphological and molecular (ITS, AFLPs) evidence. Nord J Bot
22: 129–169.
Laureau C, Meyer S, Baudin X, Huignard C, Streb P. 2015. In vivo epidermal
UV-A absorbance is induced by sunlight and protects Soldanella alpina
leaves from photoinhibition. Funct Plant Biol 42: 599–608.
Luft G. 1926. The distribution of saponins and tannins in the plant. Monatsh
Chem 47: 259–284.
Zuo Y, Zhou S, Zuo R, Shi T, Yang Y, Henegan P. 2014. Hydrophilic
interaction liquid chromatography: Fundamentals and applications. In
High-performance Liquid Chromatography, Zuo Y (ed.). Nova Science:
Hauppauge; 1–21.
Lütz C, Gülz PG. 1985. Comparative analysis of epicuticular waxes from
some high alpine plant species. Z Naturforsch C 40: 599–605.
Streb P, Aubert S, Bligny R. 2003a. High temperature effects on light
sensitivity in the two high mountain plant species Soldanella alpina
(L.) and Ranunculus glacialis (L.). Plant Biology 5: 432–440.
Streb P, Aubert S, Gout E, Bligny R. 2003b. Reversibility of cold- and light-
stress tolerance and accompanying changes of metabolite and
Supporting information
Additional Supporting Information may be found online in the
supporting information tab for this article.
wileyonlinelibrary.com/journal/pca
Copyright © 2017 John Wiley & Sons, Ltd.
Phytochem. Anal. 2017