J. Agric. Food Chem. 1996, 44, 2705?2709
Improved High-Performance Liquid Chromatographic Method for
the Analysis of Potato (Solanum tuberosum) Glycoalkaloids
Everard J. Edwards and Andrew H. Cobb*
Department of Life Sciences, The Nottingham Trent University, Clifton Lane,
Nottingham NG11 8NS, United Kingdom
An improved high-performance liquid chromatographic (HPLC) method for the determination of
the potato glycoalkaloids, -solanine and -chaconine, is described. The glycoalkaloids were extracted
with aqueous acetic acid, concentrated, and cleaned up by solid phase extraction using C18 cartridge
columns and separated using a reverse phase C18 HPLC column. Recoveries averaged 93%
for -solanine and 99%
3.1 for -chaconine. Comparisons were made at each step with existing
published methods. We conclude that trifunctional endcapped C18 SPE cartridges were more reliable
than traditional monofunctional non-endcapped cartridges and that low carbon load C18 HPLC
columns provided an improved separation of glycoalkaloids when compared to columns with a high
carbon loading. An acidic extraction medium excluding organic solvents is recommended.
-Chaconine; glycoalkaloids; HPLC; potato;
-solanine; Solanum tuberosum
tuber TGA. Modern commercial practice also neces-
sitates that these methods be rapid and inexpensive.
Glycoalkaloids are naturally occurring toxins found
There are many methods for TGA analysis reported in
in all parts of the potato plant (Solanum tuberosum L.).
the literature including mass spectrometry (Chen et al.,
The principal glycoalkaloids in potato tubers are -sol-
1994), isotachyphoresis (Kvasnicka et al., 1994), thin
-chaconine, both glucosylated forms of the
layer chromatographic scanning (Ferreira et al., 1993),
steroidal alkaloid, solanidine (Figure 1). Together they
comprise approximately 95% of total potato glyco-
various colorimetic methods, gas chromatography (Law-
alkaloids (TGA) (Olsson, 1989).
son et al., 1992), countercurrent chromatography (Fuku-
hara and Kubo, 1991), high-pressure liquid chromatog-
Potato tubers typically contain about 20 60 mg of
TGA 100 g 1
raphy (HPLC), and enzyme immunoassays.
freeze-dried matter (FDM) (Griffiths et al.,
1994), equivalent to 4 12 mg of TGA 100 g 1
method has relative advantages and disadvantages,
weight (fwt). At these concentrations glycoalkaloids are
such as lack of sensitivity, a need for derivitization,
considered to enhance potato flavor.
expensive chemicals, or excessively long preparatory
concentrations in excess of 20 mg 100 g 1
steps. Currently the most realistic methods utilized are
impart a bitter taste and can cause gastroenteritic
HPLC and immunoassays. The latter offers the pos-
symptoms, coma, and even death. The toxic dose is
sibility of a sensitive, simple, rapid, and relatively cheap
considered to be approximately 2 5 mg kg 1
detection method. However, polyclonal antibodies re-
whereas the lethal dose is probably only 3 6 mg kg 1
ported in the literature show a lack of specificity or
(Morris and Lee, 1984).
immunological response (Morgan et al., 1983; Plhak and
Approximately 30 deaths, more than 2000 cases of
Sporns, 1992; Plhak and Sporns, 1994) and the mono-
poisoning, and numerous livestock losses have been
clonals that have been raised to date, although specific
reported as caused by consumption of potatoes with high
to solanidine-based compounds, cannot distinguish be-
concentrations of glycoalkaloids (Morris and Lee, 1984).
tween individual glycoalkaloids, making them unsuit-
Indeed, as recently as 1993 a Japanese woman was
able for some studies (Plhak and Sporns, 1994; Stanker
taken ill after using a folk remedy prepared from
et al., 1994).
potatoes containing 15-fold normal TGA content (Gono-
There are a great many published HPLC methods due
mori et al., 1993).
to the proven ability of HPLC to separate and quantify
The concentration of tuber glycoalkaloids increases
potato glycoalkaloids. In addition, there is a wide
in response to a number of factors, including physical
availability of HPLC equipment and these methods have
injury (either from harvesting and sorting or herbivory),
fungal attack, poor growing conditions, climate, and,
the advantage of speed and ease of use. The majority
perhaps most importantly, inadequate storage condi-
of reported HPLC methods use reverse phase (RP) C18
tions. It has long been known that exposure of tubers
or NH2 sorbents with a mobile phase consisting of
to light can rapidly cause a large increase in TGA
acetonitrile and a biological buffer, commonly phos-
concentrations (Conner, 1937). This can occur during
phate. However, extraction techniques and reported
growth, harvesting, transport, storage, and at point of
recoveries vary greatly. Indeed, with over 400 papers
sale. Therefore light exposure is probably the most
published in the past 25 years (Plhak and Sporns, 1994)
important commercial factor influencing TGA content
there is much contradiction and confusion in the litera-
in potato tubers. Breeding with wild Solanum biotypes
ture on TGA analysis. The main objective of this study
for new varieties can also result in unacceptably high
has been to rationalize and consolidate these methods
TGA levels (Sanford et al., 1992).
and produce a concise, reproducible method for the
Consideration of the above demonstrates the impor-
routine extraction and quantification of potato glyco-
tance of accurate and reliable methods of analysis for
S0021-8561(95)00740-0 CCC: $12.00
© 1996 American Chemical Society
2706 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Edwards and Cobb
Figure 1. Chemical structures of the major potato alkaloids.
hundred milligrams of the powder was mixed with 15 mL of
The extract was homogenized with an Ultra Turrax TP18/
10 with a 10N probe (Fisons, Loughborough, UK), for 30 s and
then centrifuged at 5310g in a MSE Chilspin (Fisons) at 4 °C
for 15 min. Samples were kept on ice throughout the proce-
dure. All solutions were prepared with ultrapure water from
an Elga Maxima (Elga Ltd., High Wycombe, UK) and all
chemicals were AnalaR grade unless otherwise stated.
Purification and Concentration. Ten milliliters of the
supernatant was applied to a solid phase extraction (SPE)
column. A comparison was made between four commercially
available SPE sorbents; Sep-Pak C18 (Waters, Watford, UK),
Isolute monofunctional non-endcapped C18 (MF-NE), Isolute
trifunctional non-endcapped C18 (TF-NE), and Isolute trifunc-
tional endcapped C18 (TF-EC) (all from Jones Chromatography,
Hengoed, UK). Four weights of the TF-EC sorbent, 200, 500,
1000, and 2000 mg, were tested to determine optimal recover-
ies of spiked samples.
The columns were activated with 5 10 mL of methanol,
according to sorbent volume, and equilibrated with 10 mL of
extraction medium. Potato extract was applied to the column
and interfering constituents were removed with a protocol
adapted from Friedman and Levin (1992). This consisted of
a series of washes: 5 mL of water, 5 mL of 0.05 M ammonium
bicarbonate, 5 mL of methanol ammonium bicarbonate (50:
50 v/v), and 5 mL of water. All washings were collected and
tested for the presence of glycoalkaloids. The glycoalkaloids
were finally eluted with 4 8 mL of methanol 0.1 M hydro-
chloric acid (80:20 v/v). The eluent was neutralized with 15
µL of 2 M NaOH/mL and dried under vacuum with a Jouan
RC 10.22 centrifugal evaporator (Tring, UK). The residue was
resuspended with 1 mL of methanol 0.5 M HCl (60:40 v/v)
and 50 µL samples were analyzed by HPLC.
Separation and Analysis. HPLC of TGA was performed
Figure 2. Flow diagram of method described in the text.
using a System Gold 126 pump, 166 variable wavelength UV
MATERIALS AND METHODS
detector, and 507 autosampler (all Beckman Instruments,
High Wycombe, UK) controlled with a Viglen SL personal
computer (Alperton, UK) using System Gold chromatography
Plant Material. Potato tubers (cv. King Edward) were
software (Beckman Instruments).
bought from a local supermarket, washed, and used on the
day of purchase. Following a detailed evaluation and verifica-
Eight RP columns, six C18 and two C8, were evaluated (Table
tion of the available published methods, the scheme illustrated
1) using a method adapted from that of Jonker et al. (1992).
in Figure 2 was adopted for all further studies. Each step is
The mobile phase was acetonitrile 0.01 M Tris HCl buffer (40:
now described in detail.
60 v/v) adjusted to pH 7.8 with HCl. This was filtered through
TGA Extraction. A 10 mm diameter cork borer was used
a 0.22 µm filter and degassed for 20 min under reduced
to take samples vertically through fresh tubers, avoiding the
pressure. The HPLC flow rate was 1.5 mL min 1
, and the
apex, base, and eyes. These were cut in half and either (a)
detector was set at 202 nm. The method was calibrated using
ground fresh in 15 mL of extraction medium (0.02 M heptane-
-chaconine standards (purity
sulfonic acid in 1% aqueous acetic acid (v/v) with 1 mg/mL
Gillingham, UK) dissolved in methanol 0.5 M HCl (60:40 v/v).
sodium bisulfite) using a mortar and pestle or (b) frozen in 10
Recovery was estimated using internal standards of 0.25 mg
mL of 0.1% aqueous poly(vinylpyrrolidone) (w/v) to
per sample with four spiked and four control samples taken
lyophilized, and ground to a powder with a glass rod. Five
from the same tuber.
HPLC Method for Analysis of Potato Glycoalkaloids
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2707
Table 1. HPLC Columns Used
particle size (µm)
HPLC Technology, Macclesfield, UK
HPLC Technology, UK
HPLC Technology, UK
Beckman Instruments, UK
Table 2. SPE Sorbent Type Recoveriesa
Table 4. Separation, Peak Shape, and Retention of
Glycoalkaloids by Representative C
mean recovery (%)
8 and C18 HPLC
Techsphere 80 C
a Results are the mean percentage recovery of standards
standard errors, where n
a Separation calculated as Rs
2), where Rs
Table 3. SPE Sorbent Volume Recoveriesa
retention time, and w
peak width at baseline;
peak asymmetry calculated as A
b/a, where A
mean recovery (%)
rear baseline segment at 10% peak height, and a
sorbent weight (mg)
baseline segment at 10% peak height. Zorbax C8 results with a
mobile phase of 55:45 Tris:acetonitrile.
a Sorbent used was TF-EC. Results are the mean percentage
recovery of standards from spiked samples
The use of freeze-dried powder gave recoveries with
no significant difference from those from fresh tissue
(data not shown). The highest recovery of 0.5 mg/mL
standards was achieved using Isolute TF-EC columns
(Table 2), with virtually 100% recovery, whereas Sep-
Pak columns gave only 25 30%. None of the other SPE
columns used gave recoveries as high as those obtained
with the TF-EC.
Initially, 200 mg columns were used but it was found
that breakthrough of sample occurred for potato samples
spiked with standards. It was, therefore, necessary to
increase the amount of sorbent in the columns. One
gram of sorbent weight was found to give the best
recovery (Table 3). It is probable that the 2 g columns
gave a lower recovery due to irreversible adsorption of
Glycoalkaloids were not detected in any of the washes
used to elute unwanted components from the SPE
The degree of separation of
nine by HPLC was dependent on both column choice
and the composition of the mobile phase. Two sorbents,
µBondapak and Techsphere BDS, gave no separation
of the standards. Zorbax C18 produced uneven peaks
with excessively long retention times.
columns gave greater than 90% separation with 0.5 mg/
Figure 3. Chromatogram of 0.25 mg/mL standards.
mL standards: Techsphere 80 C18 and the two C8
sorbents (Table 4). Asymmetry of the eluted peaks
varied greatly between the columns, with the C8 sor-
When analyzing potato tubers for TGA, it is important
bents giving the least tailing and Spherisorb the most.
to consider the sampling procedure. Almost all TGA
Techsphere 80 was subject to the least band broadening,
found in freshly harvested, nongreened, potatoes are
had short retention times, and was used for all subse-
found in the outer layers (Kozukue et al., 1987), and
quent work (Figure 3).
small and immature tubers have greater TGA concen-
2708 J. Agric. Food Chem., Vol. 44, No. 9, 1996
Edwards and Cobb
trations (Hellenas, 1995). The results obtained by any
and -chaconine. A number of published methods have
sampling method are consequently affected by the ratio
used acid phosphate buffers (Hellenas, 1986). However,
of the surface area of peel to volume of sample. The
full separation could not be attained with the HPLC
effect of this is greatest when using whole tubers but
columns reported above. Jonker et al. (1992) proposed
can be minimized by using half cores taken through the
the use of Tris HCl at near-neutral pH. Good separa-
center of the potato.
tion was achieved on the Techsphere 80 column using
Since tuber tissues are approximately 80% water by
this mobile phase.
weight, the freeze-drying of samples provides a 5-fold
Separation and retention of the analytes was in-
concentration of TGA. It also allows samples to be
creased by raising the buffer pH. However, this resulted
stored for longer periods of time prior to analysis. In
in reduced sensitivity of the system due to the glyco-
this study, no differences were found in TGA content
alkaloids being less soluble. No TGA was detected
between fresh and lyophilized potato samples. Consid-
above pH 8 due to precipitation upon injection onto the
ering these advantages, freeze-dried tissues were used
column. The optimum pH was found to be 7.6 7.8.
throughout this research project.
Reducing the amount of acetonitrile in the mobile phase
There is a wide range of extraction solvents employed
also improved separation. Using 40% acetonitrile, 60%
in published methods. Most are based on a weak
buffer gave maximum separation, although glycoalka-
solution of acetic acid with the addition of other solvents
loids are insoluble at lower acetonitrile ratios.
or salts. Methanol (Jonker et al., 1992) or tetrahydro-
Detection of potato glycoalkaloids requires the use of
furan (THF) (Bushway et al., 1986) is commonly added.
UV as there is no chromophore within the solanidine
However, as glycoalkaloids are probably stored within
molecule. This limits sensitivity of detection, requiring
the aqueous phase of the potato cell and are readily
the use of relatively high sample volumes. Spectra of
soluble in dilute acid, the use of these solvents is
-solanine and -chaconine in the adopted mobile phase
unnecessary. Furthermore, they require removal before
show maxima at 202 and 203 nm, respectively. There-
any SPE step as many organic solvents prevent full
fore, 202 nm was used for quantification of TGA. This
adsorption of TGA onto C
also means that mobile phases containing THF are
18 SPE sorbents.
ditional steps required for the removal of such solvents
unsuitable as it has an UV cutoff of 212 nm.
further contribute to a reduced recovery. The use of
The method described in this paper addresses a
heptanesulfonic acid as an ion-pair reagent (Carmen et
number of problems with HPLC analysis of potato
al., 1986) enhances complete adsorption of TGA and was
glycoalkaloids and utilizes a new SPE sorbent to im-
therefore used in these investigations. Sodium bisulfite
prove sample recoveries resulting in a reliable routine
was used to reduce oxidation of the extract (Hellenas,
assay of use to both researchers and the potato industry.
It is being used in this laboratory to investigate the
A number of published methods involve an extraction
response of potato tuber TGA concentrations to envi-
step using methanol (e.g., Kvasnicka, 1994). However,
ronmental conditions during storage.
the present study found potato glycoalkaloids to be
virtually insoluble in cold methanol and so recovery
from these methods can be at best uncertain, even at
the low concentrations present in most potato tubers.
Bushway, R. J. M.; Bureau, J. L.; King, J. Modification of the
rapid high-performance liquid chromatographic method for
Virtually every recent paper describing a clean-up
the determination of potato glycoalkaloids. J. Agric. Food
method for TGA advocates the use of Sep-Pak C18 SPE
Chem. 1986, 34, 277 279.
cartridges. However, we have observed extremely vari-
Carman, A. S.; Kuan, S. S.; Francis, O. J.; Kirschenheuter, G.
able results between batches and tested three other
P. Rapid HPLC determination of the potato glycoalkaloids
cartridge types for reproducibility of recoveries. The
alpha-solanine and alpha-chaconine. J. Agric. Food Chem.
Isolute sorbent equivalent to Sep-Pak C18 (MF-NE) gave
1986, 34, 279 282.
recoveries similar to Sep-Pak, but the TF-EC sorbent
Chen, S.; Derrick, P. J.; Mellon, F. A.; Price, K. R. Analysis of
gave the highest recovery and good reproducibility.
glycoalkaloids from potato shoots and tomatoes by four-
Unfortunately, the low selectivity of endcapped C
sector tandem mass spectrometry with scanning array
sorbents necessitated a large sorbent weight due to the
detection: comparison of positive ion and negative ion
high number of compounds in the extract capable of
methods. Anal. Biochem. 1994, 218, 157 169.
binding to the column. No significant batch to batch
Conner, H. W. Effect of light on solanine synthesis in the
potato tuber. Plant Physiol. Lancaster 1937, 12, 79 98.
variation has yet been observed with these cartridges,
Ferreira, F.; Moyna, P.; Soule, S.; Vazquez, A. Rapid deter-
and therefore, these cartridges have been adopted for
mination of solanum glycoalkaloids by thin-layer chromato-
all subsequent work.
graphic scanning. J. Chromatogr. 1993, 53, 380 384.
While Friedman and Levin (1992) did not detect any
Friedman, M.; Levin, C. E. Reversed phase HPLC separation
hydrolysis of TGA when evaporating to dryness in the
of potato glycoalkaloids and their hydrolysis products on
presence of acid, this investigation found that drying
acidic columns. J. Agric. Food Chem. 1992, 40, 2157 2163.
samples in dilute HCl gave extremely variable recover-
Friedman, M.; McDonald, G. M. Acid catalyzed partial hy-
ies. Friedman et al. (1993, 1995) demonstrated that
drolysis of carbohydrate groups of the potato glycoalkaloid
50% of -chaconine is hydrolyzed in 0.2 N HCl at 50 °C
-chaconine in alcoholic solutions. J. Agric. Food Chem.
and that the presence of methanol increases the rate of
1995, 43, 1501 1506.
hydrolysis. Neutralizing the SPE eluate prevented this
Friedman, M.; McDonald, G.; Haddon, W. F. Kinetics of acid-
hydrolysis, and the excess NaOH had no detectable
catalyzed hydrolysis of carbohydrate groups of potato gly-
-solanine. J. Agric. Food
adverse effect on the glycoalkaloid extract. The surplus
Chem. 1993, 41, 1397 1406.
alkali did, however, make it necessary to resuspend the
Fukuhara, K.; Kubo, I. Isolation of steroidal glycoalkaloids
samples in a stronger acid than was used for the SPE
from Solanum incanum by two countercurrent chromato-
graphic methods. Phytochemistry 1991, 30, 685 687.
The composition of the mobile phase used for HPLC
Gonomori, K.; Meguro, H.; Lu, Y. Q.; Yoshioka, N.; Hori, K.;
was important in ensuring full separation of -solanine
Kikkawa, S. The risk of solanine poisoning in a folk remedy
HPLC Method for Analysis of Potato Glycoalkaloids
J. Agric. Food Chem., Vol. 44, No. 9, 1996 2709
and solanine production in potato. Res. Pract. Forensic Med.
Olsson, K. Anatomical structure and chemical composition of
1993, 36, 91 95.
potato tubers. In: Impact Damage, Gangrene and Dry Rot
Griffiths, D. W.; Dale, M. F. B.; Bain,H. The effect of cultivar,
in Potato, Important Biochemical Factors in Screening for
maturity and storage on photo-induced changes in the total
Resistance and Quality in Breeding Material; Olsson, K.,
glycoalkaloid and chlorophyll contents of potatoes (Solanum
Ed.; Swedish University of Agricultural Sciences: Uppsala,
tuberosum). Plant Sci. 1994, 98, 103 109.
Sweden, 1989; pp 8 21.
Hellenas, K. A simplified procedure for the quantification of
Plhak, L. C.; Sporns, P. Enzyme immunoassay for potato
potato glycoalkaloids in tuber extracts by HPLC; comparison
glycoalkaloids. J. Agric. Food Chem. 1992, 40, 2533 2540.
with ELISA and a colorimetric method. J. Sci. Food Agric.
Plhak, L.; Sporns, P. Development and production of mono-
1986, 37, 776 782.
clonal antibodies for the measurement of solanidine potato
Hellenas, K. Glycoalkaloid content of early potato varieties.
glycoalkaloids. Am. Potato J. 1994, 71, 297 313.
J. Sci. Food Agric. 1995, 67, 125 128.
Jonker, H. H.; Koops, A. J.; Hoogendoorn, J. C. A rapid method
Sanford, L. L.; Deahl, K. L.; Sinden, S. L.; Ladd, T. L.
for the quantification of steroidal glycoalkaloids by reversed
Glycoalkaloid contents in tubers from Solanum tuberosum
phase HPLC. Potato Res. 1992, 35, 451 455
populations selected for potato leafhopper resistance. Am.
Kozukue, N.; Kozukue, E.; Mizuno, S. Glycoalkaloids in potato
Potato J. 1992, 69, 693 703.
plants and tubers. HortScience 1987, 22, 294 296.
Stanker, L. H.; Kamps-Holtzapple, C.; Friedman, M. Develop-
Kvasnicka, F.; Price, K. R.; Ng, K.; Fenwick, G. R. Determi-
ment and characterization of monoclonal antibodies that
nation of potato glycoalkaloids using isotachophoresis and
differentiate between potato and tomato glycoalkaloids and
comparison with a HPLC method. J. Liq. Chromatogr. 1994,
aglycones. J. Agric. Food Chem. 1994, 42, 2360 2366.
17, 1941 1951.
Lawson, D. R.; Erb, W. A.; Millar, A. R. Analysis of Solanum
alkaloids using internal standardization and capillary gas
Received for review November 8, 1995. Revised manuscript
chromatography. J. Agric. Food Chem. 1992, 40, 2186 2191.
received May 30, 1996. Accepted June 7, 1996.X This study
Morgan, M. R. A.; McNeney, R.; Matthew, J.; Coxon, D. T.;
was partially funded by the Potato Marketing Board of the
Chan, H. W. S. An enzyme linked immunosorbent assay for
United Kingdom and The Nottingham Trent University.
total glycoalkaloids in potato tubers. J. Agric. Food Chem.
1983, 34, 593 598.
Morris, S. C.; Lee, T. H. The toxicity and teratogenicity of
Solanaceae glycoalkaloids, particularly those of the potato
(Solanum tuberosum): a review. Food Technol. Aust. 1984,
X Abstract published in Advance ACS Abstracts, Au-
36, 118 124.
gust 1, 1996.