J. Sep. Sci. 2008, 31, 37273731 C. L. Copper et al.

3727
Christine L. Copper
1
Carl I. D. Newman
2
Greg E. Collins
2
1
United States Naval Academy,
Chemistry Department,
Annapolis, MD, USA
2
Naval Research Laboratory,
Chemistry Division,
Washington, DC, USA
Original Paper
Simple and rapid extraction, separation, and
detection of alkaloids in beverages
Implementation of an uncomplicated SPE process for the rapid extraction and pre-
concentration of the alkaloids, colchicine, strychnine, aconitine, and nicotine, from
water, apple juice, and nonfat milk samples is presented. When coupled to analysis
via micellar EKC (MEKC), the total analysis time per sample was less than 15 min for
the water and juice samples and less than 20 min for the milk. The SPE process
allowed for anywhere from a three to a fourteen-fold improvement in the LOD for
each alkaloid when compared to detecting the alkaloids in a nontreated water sam-
ple matrix. Following SPE, the LODs for colchicine, strychnine, and nicotine were
sufficient to meet levels from 150 to 5000 times more dilute than the LD
50
for a
50 kg individual drinking 12 oz of a contaminated beverage. Aconitine, on the
other hand, was detected at approximately the LD
50
level. The percent recoveries for
the SPE ranged from 37% to as high as 99%. Nicotine attained the highest recovery
efficiencies, followed by colchicine, and finally, aconitine and strychnine, which
were nearly identical. The greatest recovery efficiencies were achieved from apple
juice and water, whereas nonfat milk yielded the lowest.
Keywords: Alkaloids / Micellar electrokinetic chromatography / SPE /
Received: June 17, 2008; revised: August 7, 2008; accepted: August 8, 2008
DOI 10.1002/jssc.200800350
1 Introduction
Alkaloids are organic, nitrogen-based compounds known
for their medicinal and, paradoxically, poisonous attrib-
utes. Alkaloids such as aconitine, colchicine, nicotine,
and strychnine are derived from plants and are easily
obtainable (see chemical structures in Fig. 1). These alka-
loids can be introduced to food products innocuously
(e.g., in milk, via cattle and sheep grazing on alkaloid-pro-
ducing plants, or veterinary use of colchicine on cattle)
[1] or insidiously (e.g., intentional poisoning) [2, 3]. The
low LD
50
of most of the alkaloids and their solubility in
many solvents makes thempotential poisoning agents of
choice. These attributes have placed colchicine, nicotine,
and strychnine on the US Centers for Disease Control
and Prevention's list of biotoxins of concern in case of a
chemical emergency [http://www.bt.cdc.gov/agent/agent
listchem-category.asp]. This classification for these alka-
loids has increased the interest in developing rapid,
selective, and sensitive alkaloid detection methods.
In the event of a compromise of the food supply or for
broad, preventative food screening of large quantities of
samples, analysis time and sensitivity are parameters of
great importance. The capability to rapidly screen sam-
ples to ascertain the presence of certain alkaloids at
Correspondence: Dr. Greg E. Collins, Naval Research Labora-
tory, 4555 Overlook Avenue, S.W., Chemistry Division, Code
6112, Washington, DC 20375-5342, USA
E-mail: greg.collins@nrl.navy.mil
Fax: +1-202-404-8119
Abbreviations: MEKC, micellar EKC; PF, preconcentration factor
i 2008 WILEY-VCHVerlag GmbH&Co. KGaA, Weinheim www.jss-journal.com
Figure 1. Chemical structures for the alkaloids examined in
this study: aconitine, nicotine, strychnine, and colchicine.
3728 C. L. Copper et al. J. Sep. Sci. 2008, 31, 37273731
some predetermined threshold concentration (i.e., LD
50
)
is of utmost importance. If a rapid food screening tech-
nique should alert to a given food or beverage as poten-
tially containing harmful levels of alkaloid toxins, more
extensive analysis techniques can then be employed to
determine the exact nature of this contaminant.
Due to the complex matrices associated with food sam-
ples, analysis is dependent upon an effective extraction
method for withdrawing and preconcentrating the ana-
lyte of interest from an otherwise complex and difficult
background. Of the various extraction techniques avail-
able, SPE is one of the most convenient because it lends
itself naturally to field analyses and is efficient with
respect to time, cost, and resources, particularly for bev-
erage samples [4]. The wide variety of SPE packings and
options for elution solvents facilitate the application of
numerous elution mechanisms for contaminant
removal.
Previous efforts to analyze alkaloids in beverages have
included an SPE step, but consisted of complex and time-
consuming multistep procedures. Smallwood et al. [5]
presented a method for quantifying nicotine, colchicine,
strychnine, and aconitine, among other compounds, in
3% milk, chocolate milk, orange juice, and blended vege-
table juice. They report a process by which an ion-pairing
reagent is added to a spiked beverage sample and then
centrifuged. The supernatant is filtered and processed
using a C
18
SPE column prior to HPLC analysis. This
method is reported to have recoveries of 90% nicotine
and 65% strychnine, as well as LODs of 3 lg/mL (19 lM)
for nicotine and 5 lg/mL (15 lM) for strychnine. The anal-
ysis time for each sample was on the order of 55 min per
sample; much of which is required for centrifuging and
a lengthy HPLC retention time for aconitine. Jablonski et
al. [6] reported an analysis of dairy products for nicotine,
strychnine, and aconitine using liquidliquid extraction
followed by SPE to give a sample that is also analyzed by
HPLC. This method is reported to achieve extraction effi-
ciencies between 72 and 89%for these analytes fromnon-
fat and whole milk and cream samples. No LODs were
reported due to the stated need for more recoveries and
controls, although the samples analyzed were in the
microgram per milliliter concentration range. The anal-
ysis time of this method was close to 75 min for each
milk sample and 85 min or more for each cream sample,
due to an additional heating step in the latter. A signifi-
cant portion of the analysis time was due to a lengthy
drying step in the SPE process that was not performed by
Smallwood or by our group.
In order to improve upon the speed of analysis, we dis-
cuss here the implementation of an uncomplicated SPE
process for rapid extraction and preconcentration of col-
chicine, strychnine, aconitine, and nicotine from water,
apple juice, and nonfat milk samples. When coupled to
analysis via micellar EKC (MEKC), the total analysis time
per sample including SPE is less than 15 min for the
water and juice samples and less than 20 min for the
milk; an improvement of three to four times over the cur-
rent state-of-the-art. Our interest in MEKC stems from the
advantage of shorter analysis times, smaller sample
requirements, and more efficient separations when com-
pared to HPLC [7]. In addition, our group has demon-
strated the rapid separation of alkaloid mixtures utiliz-
ing a lab-on-a-chip platform with MEKC-based separation
and spectrographic UV-absorbance detection, indicating
the potential for portable sensing of alkaloids in the
future [8, 9].
2 Materials and methods
2.1 Chemicals and reagents
Apple juice and nonfat milk samples were purchased
from a local grocery store and used prior to their expira-
tion dates. All other chemicals were purchased from
SigmaAldrich (St. Louis, MO). Spec DAU 3 mL SPE col-
umns were purchased from Varian (Lake Forest, CA).
Fresh BGE solutions of 80 mM sodium cholate and
10 mM Tris were made daily to an adjusted pH of 9.0
from concentrated stock solutions. All MEKC separations
were performed using this cholate/Tris BGE due to pre-
vious results which indicated that this matrix provides
maximal resolution and sensitivity [9]. Stock solutions of
aconitine, colchicine, nicotine, and strychnine were pre-
pared in ethanol at concentrations of 7743, 2679, 8938,
and 3977 lM, respectively. Standard solutions were
made by pipetting appropriate amounts of each alkaloid
stock solution into a vial, evaporating to dryness with
compressed air, and reconstituting each vial with ali-
quots of BGE to yield samples from200 to 1000 lM.
2.2 Apparatus and instrumentation
Separations were performed with a Beckman-Coulter P/
ACE
TM
MDQ Series Capillary Electrophoresis System (Full-
erton, CA) in a 30 cm (256365 lM) fused-silica capillary
column (Polymicro Technologies, Phoenix, AZ). The col-
umn was rinsed (60 s at 10 psi) prior to each injection
with 0.1 M NaOH, water, and then BGE. Samples were
then injected hydrostatically (3 s at 0.5 psi) prior to sep-
aration under an applied electric field of 567 V/cm
(17 kV) which produced 13 lA of current. On-column UV-
absorbance detection at 248 and 232 nm was performed
20 cm from the inlet. These wavelengths facilitated max-
imum S/N detection for colchicine, nicotine, and strych-
nine at 248 nm, as well as aconitine at 232 nm.
i 2008 WILEY-VCHVerlag GmbH& Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2008, 31, 37273731 Electrodriven Separation 3729
2.3 Sample preparation
Water, apple juice, and nonfat milk samples (2 mL each)
were spiked to a concentration varying from 10 to 50 lM
in each alkaloid. Milk proteins were precipitated by the
addition of 0.5 mL of 5 M HNO
3
, and centrifuged
(RPM = 1550) for 5 min. An in-house built manifold and
vacuum pump was used to facilitate flux through the
SPE columns during rinsing and sample loading proce-
dures. The columns were activated with 1 mL of metha-
nol and rinsed with 1 mL of 5 M HNO
3
, prior to addition
of the spiked samples. Sample vials were then rinsed
with 0.5 mL 5 M HNO
3
and the rinse added to the SPE col-
umn. The columns were dried under vacuum for 5 min
and then removed from the manifold. One milliliter of
elution solvent (78:20:1, ethyl acetate/methanol/conc.
NH
4
OH) was added to the SPE column and forced
through with compressed air. The eluent was evaporated
to dryness and then reconstituted in 0.1 mL of BGE. Thus,
assuming 100% extraction and recovery, the potential
for a 20-fold concentration enhancement exists.
2.4 Safety
Due to their acute toxicity, neat alkaloids should be
handled with care using proper ventilation and appro-
priate gloves. Alkaloid solutions should also be respected
and handled with gloved hands. All solutions should be
disposed of properly.
3 Results and discussion
Figure 2 shows an electropherogram of a 1000 lM stand-
ard solution. The full electropherogram in the figure
shows the UV absorbance at 248 nm, whereas the inset
peaks represent UV absorbance from aconitine and
strychnine at 232 nm. Note that aconitine and strych-
nine are baseline resolved under these separation condi-
tions. By comparison, the aconitine signal is much larger
in the 232 nm trace, while the strychnine peak is much
smaller, due to differences in their maximum absorban-
cies. The significant advantage of MEKC analysis is dem-
onstrated by the rapid analysis time of 2.5 min.
Electropherograms of five calibration standards from
200 to 1000 lMin water were collected at 248 nm. Linear
regression analysis of peak areas over this concentration
range was performed on the 248 nm data for nicotine,
strychnine, and colchicine; data collected at 232 nm
were utilized for the aconitine. The R
2
values for nicotine,
aconitine, strychnine, and colchicine were 0.998, 0.979,
0.998, and 0.996, respectively. These calibration data
were later used to quantify the concentration of each
alkaloid in the spiked water, apple juice, and nonfat
milk samples.
Table 1 summarizes the analytical parameters for the
MEKC analysis of these four alkaloids in BGE in the
absence of SPE. Also indicated in this table is the LD
50
i 2008 WILEY-VCHVerlag GmbH&Co. KGaA, Weinheim www.jss-journal.com
Table 1. Analytical parameters for the MEKC analysis of nicotine, aconitine, strychnine, and colchicine in BGE
Alkaloid toxin LD
50
(rat)
(mg/kg)
Lethal dose for
50 kg person
k
(nm)
Sensitivity parameters for no SPE in BGE for 2001000 lM
(lM/12 oz) R
2
N RSD(%) LOD(lM)
Nicotine 50 43 529 248 0.998 12 3 97
Aconitine 0.08 20 232 0.979 12 10 94
Strychnine 2.35 990 248 0.998 12 3 37
Colchicine 1.6 570 248 0.996 12 11 15
Coefficient of determination = R
2
, sample set size = N.
Figure 2. MEKC electropherogram of 1000 lM (a) standard
of nicotine, (b) aconitine, (c) strychnine, and (d) colchicine
using UV-absorbance detection at 248 nm. The inset electro-
pherogram displays the peaks obtained for (b) aconitine and
(c) strychnine using UV-absorbance detection at 232 nm.
BGE and sample matrix: 10 mM Tris base, 80 mM sodium
cholate at pH 9.0. Hydrostatic injection: 3 s at 0.5 psi.
Applied electric field: 567 V/cm (17 kV). Typical operating
current: l13 lA.
3730 C. L. Copper et al. J. Sep. Sci. 2008, 31, 37273731
(rat) for each separate alkaloid and how that LD
50
trans-
lates to a concentration (lM) in a 12 oz beverage. Nicotine
is clearly the least toxic of the four alkaloids, requiring
A43 mM to deliver a lethal dose to a 50 kg individual;
whereas aconitine is the most toxic at just 20 lM for a
50 kg individual. Finally, Table 1 lists the calculated LOD
obtainable fromthe MEKC analysis for each of these alka-
loids in BGE in the absence of SPE. These LODs were calcu-
lated from the linear regression statistics for each alka-
loid, assuming a S/Nof three.
For the analysis of actual beverages, 2 mL samples of
water, apple juice, and nonfat milk were spiked to con-
tain from 10 to 50 lM of each alkaloid and then proc-
essed according to the details given in Section 2. The SPE
material utilized is comprised of fused-silica membranes
that are functionalized with aliphatic and strong cation
exchange moieties which attain maximal retention by
the concerted effects of partitioning and ion-exchange.
These columns have been successfully utilized for the
extraction of drugs of abuse, including the opiates, mor-
phine, and codeine, for example, and were, therefore,
estimated to be well suited for the extraction of the alka-
loids of interest here [10].
Optimization studies indicated that preparing the SPE
column prior to introduction of the sample by washing
with 5 M HNO
3
helped to significantly improve the
extraction efficiencies for these alkaloids. Lowering the
pH ensures protonation of the alkaloids and facilitates
the more selective cation-exchange processes for retain-
ing the alkaloids. Although colchicine has an especially
high pK
a
value, 12.35, the remaining three alkaloids stud-
ied have pK
a
's less than 8. The application of 5 M HNO
3
was experimentally derived based on the observation
that when water or 1 M HNO
3
was used to rinse the SPE
column before and after loading, significantly lower
recoveries were realized due to the measurable presence
of alkaloids in the rinse solution.
Figure 3 displays electropherograms obtained follow-
ing SPE from water, apple juice, and nonfat milk for bev-
erages containing aconitine, colchicine, nicotine, and
strychnine at 50 lM. Comparison of Figs. 2 and 3 shows
that peaks fromthe beverage samples exhibit asymmetry
relative to those of the standard solution. This asymme-
try likely results from matrix contributions from the SPE
process, increasing the sample zone ionic strength and
lowering the local EOF in the sample zone. This asymme-
i 2008 WILEY-VCHVerlag GmbH& Co. KGaA, Weinheim www.jss-journal.com
Table 2. Sensitivity and analytical parameters for the MEKC analysis of nicotine, aconitine, strychnine, and colchicine following
SPE extraction from water, apple juice, and nonfat milk
Alkaloid
toxin
N Sensitivity and analytical characterization parameters with SPE
Water Apple juice Nonfat milk
%R RSD
(%)
PF LOD
(lM)
%R RSD
(%)
PF LOD
(lM)
%R RSD
(%)
PF LOD
(lM)
Nicotine 15 99 16 11 8.7 90 7 12 8.4 83 15 6.9 14
Aconitine 15 60 23 6.3 15 60 14 5.5 17 39 22 2.8 33
Strychnine 15 63 6 12 3.2 58 12 7.9 4.7 37 10 5.7 6.5
Colchicine 15 76 5 14 1.1 72 10 11 1.4 45 11 6.5 2.3
Sample set size = N, percent recovery = % R.
Figure 3. MEKC electrophero-
grams of (a) nicotine, (b) aconitine,
(c) strychnine, and (d) colchicine
following extraction from water,
apple juice, and nonfat milk. The
negative peak apparent at
l1.5 min is a matrix effect associ-
ated with injection of these SPE
samples. See Fig. 2 for experi-
mental conditions.
J. Sep. Sci. 2008, 31, 37273731 Electrodriven Separation 3731
try does not prevent proper identification, and by utiliz-
ing peak areas instead of peak heights, quantification of
the alkaloids was found to be reasonably reproducible as
demonstrated by the RSDvalues shown in Table 2.
The SPE method was applied to the three beverages in
a concentration range of 1050 lM. Table 2 lists the SPE
analytical parameters for all three beverages examined,
including the recovery percents (% R), RSD, and precon-
centration factors (PF) derived from the SPE process, and
the resulting LOD calculated using the MEKC method
employed in this work. Although peak areas were used
for quantitation purposes, as mentioned earlier, the
LOD's for the SPE method were calculated from peak
height and by assuming a minimal detectable S/N of 3.
Peak heights were used in this case because LOD's cannot
be theoretically determined from peak areas. The PF is a
ratio of the LOD for samples without SPE to the LOD for
post-SPE samples. The SPE process allowed for anywhere
from a three- to a fourteen-fold improvement in the LOD
for each alkaloid when compared to detecting the alka-
loid froma nontreated water sample matrix.
The percent recoveries for the SPE ranged from as low
as 37%for strychnine in nonfat milk to as high as 99%for
nicotine in water. In general, nicotine attained the high-
est recovery efficiencies, followed by colchicine, and
finally, aconitine and strychnine, which were nearly
identical. Extraction efficiency is likely closely related to
the ionic charge and hydrophobic character of the alka-
loids introduced to the SPE column. Referring to Fig. 1,
the small size and largely aromatic character of nicotine,
when combined with its dual protonation sites, made for
a very large extraction efficiency. The greatest recoveries
were achieved from water and apple juice, whereas the
lowest recoveries were derived from nonfat milk. This is
likely due to the additional precipitation and centrifuga-
tion steps required for the analysis of milk samples.
4 Concluding remarks
The SPE procedure described herein and the accompany-
ing MEKC separation are sufficient to achieve LODs for
colchicine, strychnine, and nicotine that are 1505000
times more dilute than the LD
50
for a 50 kg individual
ingesting a contaminated 12 oz beverage regardless of
whether the beverage is water, apple juice, or nonfat
milk. However, the calculated LOD for aconitine follow-
ing SPE preconcentration is at approximately its LD
50
level, and, therefore, one might consider processing a
larger volume of beverage by SPE in order to achieve
greater preconcentration. Alternatively, several groups
have demonstrated that changes in the injected sample
matrix composition will promote online sample precon-
centration (field amplified stacking, sweeping, high salt
stacking) during the MEKC analysis step which can result
in enhancement factors from 2 to 1000 times for neutral
analytes [7, 1113]. These online preconcentration tech-
niques are easily implemented in most SPE processes
because the SPE extract can simply be reconstituted in
the appropriate sample matrix. Finally, the entire anal-
ysis from SPE to completed MEKC analysis is performed
under 15 min for water and juice samples, and under
20 min for milk samples; three to four times faster than
previously reported methods [5, 6].
The authors gratefully acknowledge financial support of this
work through a grant from the Food and Drug Administration
(FDA). Its contents are solely the responsibility of the authors and
do not necessarily represent the official views of the FDA or the
National Institutes of Health (NIH). C. C. acknowledges support
from the Naval Academy Research Council and the Office of Naval
Research.
The authors have declared no conflict of interests.
5 References
[1] Hamscher, G., Priess, B., Nau, H., Panariti, E., Anal. Chem. 2005,
77, 24212425.
[2] Dehon, B., Chagnon, J. L., Vinner, E., Pommery, J., Mathieu, D.,
Lhermitte, M., Biomed. Chromatogr. 1999, 13, 235238.
[3] Brvar, M., Ploj, T., Kozelj, G., Mozina, M., Noc, M., Bunc, M., Crit-
ical Care 2004, 8, R56R59.
[4] Wille, S. M. R., Lambert, W. E. E., Anal. Bioanal. Chem. 2007, 388,
13811391.
[5] Smallwood, A. W., Tschee, C. S., Satzger, R. D., J. Agric. Food Chem.
1997, 45, 39763979.
[6] Jablonski, J. E., Schlesser, J. E., Mariappagoudar, P., J. Agric. Food
Chem. 2006, 54, 74607465.
[7] Giordano, B. C., Newman, C. I. D., Federowicz, P. M., Collins, G.
E., Burgi, D. S., Anal. Chem. 2007, 79, 62876294.
[8] Lu, Q., Copper, C. L., Collins, G. E., Anal. Chim. Acta 2006, 572,
205211.
[9] Newman, C. I. D., Giordano, B. C., Copper, C. L., Collins, G. E., Elec-
trophoresis 2008, 29, 803810.
[10] Balkov, M. A., Habrdova, V., J. Chromatogr. B 2003, 789, 93100.
[11] Liu, Z. Y., Sam, P., Sirimanne, S. R., Mcclure, P. C., Grainger, J.,
Patterson, D. G., J. Chromatogr. A 1994, 673, 125132.
[12] Quirino, J. P., Terabe, S., Science 1998, 282, 465468.
[13] Palmer, J., Munro, N. J., Landers, J. P., Anal. Chem. 1999, 71, 1679
1687.
i 2008 WILEY-VCHVerlag GmbH&Co. KGaA, Weinheim www.jss-journal.com