Characteristic ion clusters as determinants for the identification of pyrrolizidine alkaloid N-oxides in pyrrolizidine alkaloid-containing natural products using HPLC-MS analysis

Article abstract of DOI:10.1002/jms.2969

Pyrrolizidine alkaloid (PA)-containing plants are widely distributed in the world. PAs are hepatotoxic, affecting livestock and humans. PA N-oxides are often present together with PAs in plants and also exhibit hepatotoxicity but with less potency. HPLC-MS is generally used to analyze PA-containing herbs, although PA references are unavailable in most cases. However, to date, without reference standards, HPLC-MS methodology cannot distinguish PA N-oxides from PAs because they both produce the same characteristic ions in mass spectra. In the present study, the mass spectra of 10 PA N-oxides and the corresponding PAs were systemically investigated using HPLC-MS to define the characteristic mass fragment ions specific to PAs and PA N-oxides. Mass spectra of toxic retronecine-type PA N-oxides exhibited two characteristic ion clusters at m/z 118-120 and 136-138. These ion clusters were produced by three unique fragmentation pathways of PA N-oxides and were not found in their corresponding PAs. Similarly, the nontoxic platynecine-type PA N-oxides also fragmented via three similar pathways to form two characteristic ion clusters at m/z 120-122 and 138-140. Further application of using these characteristic ion clusters allowed successful and rapid identification of PAs and PA N-oxides in two PA-containing herbal plants. Our results demonstrated, for the first time, that these characteristic ion clusters are unique determinants to discriminate PA N-oxides from PAs even without the availability of reference samples. Our findings provide a novel and specific method to differentiate PA N-oxides from PAs in PA-containing natural products, which is crucial for the assessment of their intoxication.

Full text of DOI:10.1002/jms.2969

Research Article
Received: 18 December 2011
Revised: 23 January 2012
Accepted: 3 February 2012
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI 10.1002/jms.2969
Characteristic ion clusters as determinants for
the identification of pyrrolizidine alkaloid
N-oxides in pyrrolizidine alkaloid–containing
natural products using HPLC–MS analysis
Jianqing Ruan,a,b Na Li,a,b Qingsu Xia,c Peter P. Fu,c* Shuying Peng,d
Yang Yeb,d and Ge Lina,b*
Pyrrolizidine alkaloid (PA)–containing plants are widely distributed in the world. PAs are hepatotoxic, affecting livestock and
humans. PA N-oxides are often present together with PAs in plants and also exhibit hepatotoxicity but with less potency.
HPLC–MS is generally used to analyze PA-containing herbs, although PA references are unavailable in most cases. However,
to date, without reference standards, HPLC–MS methodology cannot distinguish PA N-oxides from PAs because they both
produce the same characteristic ions in mass spectra. In the present study, the mass spectra of 10 PA N-oxides and the
corresponding PAs were systemically investigated using HPLC–MS to define the characteristic mass fragment ions specific
to PAs and PA N-oxides. Mass spectra of toxic retronecine-type PA N-oxides exhibited two characteristic ion clusters at m/z
118–120 and 136–138. These ion clusters were produced by three unique fragmentation pathways of PA N-oxides and were
not found in their corresponding PAs. Similarly, the nontoxic platynecine-type PA N-oxides also fragmented via three similar
pathways to form two characteristic ion clusters at m/z 120–122 and 138–140. Further application of using these characteristic
ion clusters allowed successful and rapid identification of PAs and PA N-oxides in two PA-containing herbal plants. Our results
demonstrated, for the first time, that these characteristic ion clusters are unique determinants to discriminate PA N-oxides
from PAs even without the availability of reference samples. Our findings provide a novel and specific method to differentiate
PA N-oxides from PAs in PA-containing natural products, which is crucial for the assessment of their intoxication. Copyright ©
2012 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.
Keywords: pyrrolizidine alkaloid N-oxides; pyrrolizidine alkaloid; HPLC–MS analysis; characteristic fragment ions; fragmentation pathway
pyrrolic metabolites that can rapidly bind to cellular proteins or
INTRODUCTION
DNA in the liver to form pyrrole–protein or pyrrole–DNA adducts
leading to hepatotoxicity or tumors, respectively.[3,4,6–10]
In the recent decades, the use of herbal products has been rapidly
growing worldwide, which makes the safety assessment of these
products more challenging than ever before.[1,2] As a result, the
safety of natural products has drawn global attention. Among the
natural toxins, pyrrolizidine alkaloids (PAs) are probably one of the
*
*
Correspondence to: Ge Lin, School of Biomedical Sciences, The Chinese
University of Hong Kong, Hong Kong, China. E-mail: linge@cuhk.edu.hk
most significant groups. More than 660 PAs have been identified in
more than 6000 plants distributed in many geographical regions
worldwide, half of which are hepatotoxic and many are tumorigenic
in humans and livestock.[3–5] Consequently, PA poisoning cases
resulting from intake of PA-containing herbal medicines or foodstuffs
accidentally contaminated with PAs frequently occur worldwide.
Generally, on the basis of the necine bases, PAs are classified into
three types: the retronecine type (including its 7-stereoisomer, the
heliotridine type), the otonecine type, and the platynecine type
(Figure 1A).[3,4] The first and second types of PAs possessing a
1,2-double bond in the unsaturated necine base are hepatotoxic
and may ultimately lead to cancer in humans, whereas the
platynecine-type PAs with a saturated necine base are generally
regarded as nontoxic PAs.[3,5] It has been well established that
the metabolism of the two toxic types of PAs is catalyzed by cyto-
chrome P450 enzymes in the liver and produces the highly reactive
Peter Fu, National Center for Toxicological Research, Jefferson, AR 72079,
USA. E-mail: peter.fu@fda.hhs.gov
Jianqing Ruan and Na Li equally contribute the work and share the first
authorship.
a School of Biomedical Sciences, The Chinese University of Hong Kong,
Hong Kong, China
b Joint Research Laboratory for Promoting Globalization of Traditional Chinese
Medicines between Shanghai Institute of Materia Medica, Chinese Academy
of Sciences and The Chinese University of Hong Kong
c
National Center for Toxicological Research, Jefferson, AR 72079, USA
d Shanghai Institute of Materia Medica, Chinese Academy of Sciences,
Shanghai 201203, China
J. Mass. Spectrom. 2012, 47, 331–337
Copyright © 2012 John Wiley & Sons, Ltd.

J. Ruan et al.
9
1
7
8
A
B
2
A
6
B
4
5
3
Figure 1. Structures of (A) the common necine bases of PAs and (B) PA N-oxides investigated in the present study.
Except for the otonecine type, in which N-oxide cannot be
formed, N-oxides of other two types of PAs also naturally occur
and often coexist with PAs in plant materials. The quantities of PA
N-oxides are found to be equal to, greater, or less than their
corresponding PAs.[11] It has been reported that retronecine-type
PA N-oxides are hepatotoxic, but with less potency than the
corresponding PAs because of the necessity for metabolic reduc-
tion to the corresponding PAs in gastrointestinal tract and/or in
the liver.[4,12] The results of human and rat liver microsomal
metabolism of riddelliine N-oxide, monocrotaline N-oxide, and
retrorsine N-oxide as well as in vivo metabolism of these three PA
N-oxides in F344 rats showed that pyrrole–DNA adducts were also
generated from these PA N-oxides but in lesser amounts than those
formed from the corresponding PAs, apparently because reduction of
PA N-oxides to the parent is not the sole metabolism pathway.[13–15]
Moreover, an in vivo study showed that intraruminal infusion of
riddelliine N-oxide alone to cattle can induce typical signs of hep-
atotoxicity similar to that caused by the original toxic plant Riddell
groundsel.[16] Therefore, in addition to the analysis of PAs, the
determination of PA N-oxides in natural products is indispensable
for the safety assessment of PA-containing products.
HPLC–MS is commonly used for qualitative and quantitative anal-
ysis of PAs and PA N-oxides in the PA-containing products and
biological samples.[17–25] Our previous study[20] demonstrated that
in the positive mode, the mass spectra of retronecine-type,
otonecine-type, and platynecine-type PAs each showed two
characteristic fragmentation ions at m/z 120 and 138, 150 and 168,
and 122 and 140, respectively. These characteristic fragment ions
can be used for the identification of different types of PAs even in
the absence of reference standards. It was recently reported that
both retronecine-type and platynecine-type PA N-oxides produced
the same characteristic fragment ions of their corresponding types
of PAs.[21,25] Consequently, a convenient mass analytical methodol-
ogy needs to be developed for distinguishing PA N-oxides from the
corresponding PAs in PA-containing herbal plants and herbal pro-
ducts. In the present study, we determine that PA N-oxides exhibit
novel and unique characteristic mass fragment ion clusters that are
not formed from the PAs. We also establish that these characteristic
ion clusters can be used to distinguish PA N-oxides from PAs present
in PA-containing plants and to identify different types of PA N-oxides.
EXPERIMENTAL
Chemicals and solvents
The reference compounds monocrotaline, retrorsine, and retrorsine
N-oxide were purchased from Aldrich Chemical Co. (Milwaukee, WI,
USA). Senecionine and seneciphylline were purchased from
Extrasynthesese (Genay, France). Lycopsamine was purchased from
PhytoLab (Dutendorfer, Germany). Platyphylline was a gift from Prof.
Hai Shen Chen from the Department of Phytochemistry, The Second-
ary Military Medical University, China. Riddelliine was obtained from
Dr Po-Chuen Chan, the National Toxicology Program (NTP). Lasiocar-
pine is gift from Dr John A. Edgar, CSIRO Livestock Industries, Austra-
lia. Heliotrine was purchased from Accurate Chemical & Scientific
Corporation (Westbury, NY, USA). All the other PA N-oxides used in
this study were synthetically prepared by N-oxidation of the
corresponding PAs with 30% hydrogen peroxide.[14,26] HPLC-grade
acetonitrile and methanol (E. Merck, Darmstadt, Germany) were used
for HPLC analyses.
HPLC–MS conditions
HPLC–MS analysis was performed on an Agilent 1200 series LC sys-
tem (Agilent Technologies, Santa Clara, CA, USA), equipped with a
binary solvent delivery system, an autosampler, connecting with
an API 2000Q-Trap mass spectrometer (AB Sciex, USA). The separa-
tion of individual compounds was achieved using Agilent Extend
C18 column (250 Â 4.6 mm, 5 mm) coupled with Extend C18 guard
column (12.5Â 4.6 mm, 5 mm). The flow rate was set at 0.8 ml/min,
and the injection volume was 10 ml. The mobile phases consisted
of 0.2% ammonia water (A) and acetonitrile (B) and a gradient elu-
tion was adopted as follows: 0–15 min, 10%–20% B; 30–40 min,
35%–35% B; 45–60 min, 35%–90% B. The mass spectrometer was
operated in positive ion mode using an electrospray ionization
wileyonlinelibrary.com/journal/jms
Copyright © 2012 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2012, 47, 331–337

Unique ion clusters to determine PA N-oxides
+MS2 (368.30)
+MS2 (352.30)
2.3e6
368.2
352.3
1
2
A
A
B
3.5e5
Retrorsine N-oxide
Retrorsine
2.0e6
1.5e6
1.0e6
3.0e5
2.5e5
118.1
2.0e5
136.1
94.1
1.5e5
95.1
138.1
154.1
1.0e5
5.0e4
0.0
5.0e5
94.1
120.1
122.3
138.2
150
220.2
324.3
340.2
350
178.1
200
246.2
250
220.2
276.3
0.0
100
150
300
400
100
200
250
m/z, Da
300
350
400
m/z, Da
+MS2 (352.30)
4.0e5
+MS2 (336.30)
336.3
352.2
1
2
B
Senecionine N-oxide
Senecionine
8.0e5
7.0e5
6.0e5
5.0e5
4.0e5
3.0e5
2.0e5
120.2
3.0e5
94.1
136.1
2.0e5
138.2
95.1
1.0e5
120.1
138.2
153.2
150
220.2
94.1
154.1
150
91.0
308.3
300
1.0e5
246.2
324.3
164.3
200
0.0
0.0
100
250
300
350
400
100
200
250
m/z, Da
350
400
m/z, Da
+MS2 (428.30)
+MS2 (412.30)
428.4
120.2
1
2
C
C
D
Lasiocarpine N-oxide
Lasiocarpine
6.0e6
5.0e6
4.0e6
3.0e6
2.0e6
1.0e6
0.0
5.0e6
4.0e6
3.0e6
254.3
2.0e6
120.3
138.2
119.2
220.3
1.0e6
410.4
400
352.3
336.4
350
238.2
250
338.4
300
m/z, Da
237.2
250
154.1
138.1
150
412.4
400
0.0
100
150
200
350
450
500
100
200
300
m/z, Da
450
500
+MS2 (316.30)
+MS2 (300.30)
6.9e5
1
2
172.2
300.3
D
3.5e6
3.0e6
2.5e6
Lycopsamine N-oxide
Lycopsamine
94.1
6.0e5
138.2
136.2
5.0e5
4.0e5
94.1
111.1
2.0e6
1.5e6
1.0e6
5.0e5
0.0
316.3
156.1
138.2
120.1
3.0e5
2.0e5
1.0e5
0.0
82.1
100
108.2
154.2
82.1
226.2
108.2
150
200
250
m/z, Da
300
350
400
100
150
200
250
300
350
400
m/z, Da
+MS2 (354.30)
4.8e6
+MS2 (338.30)
1
2
354.3
338.3
E
E
Platyphylline N-oxide
Platyphylline
2.5e7
2.0e7
1.5e7
1.0e7
4.0e6
3.0e6
2.0e6
1.0e6
140.2
122.1
122.1
140.2
310.3
300
5.0e6
93.1
138.2
150
120.1
100
240.3 292.3
153.2
150
211.2
200
290.3
300
336.3
350
156.1
256.2
250
m/z, Da
0.0
0.0
400
100
200
250
350
400
m/z, Da
Figure 2. MS2 spectra of different representative PA N-oxides and their corresponding PAs.
interface with the following working parameters: ion spray voltage,
5500 V; curtain gas, 20 psi; gas 1, 30 psi; gas 2, 70 psi; source temper-
ature, 400ꢀC. All gases used were nitrogen. Declustering potential
was set at 70V, and entrance potential was set at 10 V for all modes.
Precursor ion scan (PIS) acquisition for detecting characteristic prod-
uct ions at m/z 120 and 138 or 122 and 140 was used to identify
peaks corresponding to retronecine-type PAs and their N-oxides or
platynecine-type PAs and their N-oxides and consequently unveil
the corresponding parent ions to produce the MS2 (MS/MS) spectra.
The collision energy and collision cell exit potential were set at 30 and
2 V, respectively, for MS2 analysis.
were extracted according to our previously developed method,[27]
with minor modification. Briefly, powdered roots (0.5g) of
individual herbal plants were soaked in a 10-ml capped conical
flask with 3 ml of 2.0 % sulfuric acid for 30 min. After soaking,
the sample was extracted under supersonic washer (50 Hz) for
1 h, followed by filtration through a 0.22-mm membrane and
subject to HPLC/MS analysis.
RESULTS AND DISCUSSION
The mass spectra of retronecine-type PAs and PA N-oxides
Extraction of herbs
In the present study, nine retronecine-type PA N-oxides (Figure 1B),
which include three subtypes, diester, monoester, and nonester PA
N-oxides, were investigated. For comparison, all the corresponding
PAs were also examined in parallel except retronecine. Given the
Two PA-containing herbal plants, Gynura pseudochina and
Gynura japonica, were purchased at Nujiang in Yunnan Province,
China, in August 2010. PAs and PA N-oxides present in the plants
J. Mass. Spectrom. 2012, 47, 331–337
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms

J. Ruan et al.
Figure 3. Proposed fragmentation pathways of retronecine-type diester PA N-oxides.
mass spectra shown in Figure 2 and Supplementary Figure 1, all PAs
showed two characteristic ions at m/z 120 and 138, which are in
good agreement with our previously established mass spectromet-
ric patterns for retronecine-type of PAs.[20] For the N-oxides of dies-
ter PAs, including both open chain diester (lasiocarpine N-oxide)
and macrocycle diester (N-oxides of retrorsine, riddelliine, senecio-
nine, seneciphylline and monocrotaline) PA N-oxides, their mass
spectra all exhibited two additional characteristic ion clusters at
m/z 118–120 and 136–138 (Figure 2 and Supplementary Figure 1).
We propose the mechanism that these characteristic fragment
ion clusters are generated through three individual fragmentation
pathways (Figure 3). Pathway 1 involves the cleavage of two ester
groups at C-7 and C-9 positions to produce a fragmentation ion
at m/z 154, which is further fragmented to generate the ions at
m/z 136 and 137 through the loss of H2O and OHÁ, respectively.
Pathway 2 involves the simultaneous loss of these two ester groups
and a molecule of water to generate the ion at m/z 136 followed by
further loss of a molecule of H2O or an OHÁ group, leading to the
fragment ions at m/z 118 and 119, respectively. Pathway 3 involves
the loss of the oxygen atom attached to the nitrogen of the necine
base to form the corresponding PA. Although the relatively intensity
of the corresponding ion [M + H-O]+ for most of the PA N-oxides tested
were very low in the present study, it is a common thermal degrada-
tion pathway for N-oxides of different types of chemicals under the
conditions of mass spectrometric analysis.[28] The subsequent frag-
mentation is identical to the spectrometric pattern of the retrone-
cine-type PAs and thus form the typical characteristic ions at m/z 120
and 138.[20] Consequently, these three pathways yield the two ion
wileyonlinelibrary.com/journal/jms
Copyright © 2012 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2012, 47, 331–337

Unique ion clusters to determine PA N-oxides
TIC of +Prec (120.00)
TIC of +Prec (138.00)
A1
A2
1
1
7.6e5
7.0e5
2
2.0e6
1.5e6
6.0e5
5.0e5
4.0e5
3.0e5
2.0e5
1.0e5
0.0
2
4
1.0e6
3
5
3
6
5.0e5
0.0
4
5
6
10
20
30
40
50
10
20
30
40
50
Time, min
Time, min
+MS2 (350.30) Peak 1
2.0e6
+MS2 (352.30) Peak 2
B1
B3
B5
B2
350.2
352.3
2.2e6
2.0e6
1.5e6
1.5e6
120.1
1.0e6
1.0e6
118.1
136.1
136.1
95.1
109.1
100
138.1
5.0e5
0.0
5.0e5
220.2
154.2
150
154.2
150
322.2
324.3
167.2
246.2
250
246.3
0.0
200
300
350
400
100
200
250
m/z, Da
300
350
400
m/z, Da
+MS2 (392.30) Peak 3
+MS2 (334.30) Peak 4
1.20e6
B4
392.3
334.3
1.7e6
1.5e6
1.00e6
8.00e5
1.0e6
118.1
6.00e5
4.00e5
120.2
120.1
5.0e5
136.1
154.2
332.3
316.1
300
138.2
150
2.00e5
138.2
109.0
288.2
94.1
306.3
300
220.3
270.3
250
209.2
200
0.0
0.00
100
100
150
350
400
200
250
m/z, Da
350
400
m/z, Da
+MS2 (336.30) Peak 5
+MS2 (376.30) Peak 6
B6
376.3
336.3
1.11e6
1.00e6
5.6e5
5.0e5
8.00e5
6.00e5
4.00e5
4.0e5
3.0e5
2.0e5
316.2
120.1
120.1
138.2
150
2.00e5
94.1
1.0e5
138.2
150
308.3
300
93.1
100
103.1
0.00
100
0.0
200
250
m/z, Da
350
400
200
250
m/z, Da
300
350
400
Figure 4. Chromatograms of Gynura pseudochina extract analyzed by HPLC–MS using PIS mode, (A1–A2) total ion chromatograms at m/z 120 and 138,
respectively; (B1–B6) MS2 spectra of peaks 1–6, respectively.Peak 1, seneciphylline N-oxide; peak 2, senecionine N-oxide; peak 3, seneciphyllinine
N-oxide; peak 4, seneciphylline; peak 5, senecionine; and peak 6, seneciphyllinine.
clusters at m/z 118–120 (118 and 119 from pathway 2 and 120 from
pathway 3) and m/z 136–138 (136 and 137 from pathway 1 and 138
from pathway 3).
additional specific ion clusters at m/z 120–122 and 138–140. Sim-
ilarly to retronecine-type PA N-oxides, these two characteristic
ion clusters are also expected to be generated from the similar
three fragmentation pathways as illustrated in Supplementary
Figure 2. Therefore, the fragmentation ions at m/z 120 and 121
(pathway 2) and 122 (pathway 3) and 138 and 139 (pathway 1)
and 140 (pathway 3) formed two ion clusters at m/z 120–122
and 138–140, respectively. These two unique ion clusters all had
two mass units greater than that of retronecine-type N-oxides
and therefore can also be used to differentiate platynecine-type
PA N-oxides from retronecine-type PA N-oxides.
Furthermore, it is noteworthy that only one ion cluster at m/z
136–138 was observed in monoester (heliotrine and lycopsamine
N-oxide) and nonester (retronecine N-oxide) PA N-oxide (Figure 2
and Supplementary Figure 1), which further confirms that the
ester substitution at C-7 position is essential for the formation
of ion cluster at m/z 118–120 for the retronecine-type PAs. As a
result, the presence or absence of the ion cluster at m/z 118–120
can be used as determinants to further differentiate retronecine-
type diester PA N-oxides from nonester and monoester PA N-oxides
with the ester substituted at C-9 position.
Characterization of PAs and PA N-oxides in herbal plants
The mass spectra of platynecine-type PAs and N-oxides
The developed HPLC–MS method using PIS approach was further
applied to analyze toxic retronecine-type PAs and their PA
N-oxides in two PA-containing herbal samples, G. pseudochina
and G. japonica. The total ion chromatograms of G. pseudochina
extract at PIS mode is showed in Figure 4A. The six peaks with
the retention time at 13.4 min (peak 1), 16.9 min (peak 2), 27.1
min (peak 3), 29.8 min (peak 4), 34.0 min (peak 5), and 46.6 min
(peak 6) all generated both characteristic ions at m/z 120 and
Platyphylline was chosen as the representative platynecine-type
PA in the present study. Platyphylline with the protonated molec-
ular ion at m/z 338 ([M + H]+) exhibited two specific fragment ions
at m/z 122 and 140 (Figure 2), which have been previously
proved to be the characteristic ions for this type PAs.[20] Its
N-oxide with the protonated molecular ion at m/z 356 exhibited
not only the same two characteristic ions but also the two
J. Mass. Spectrom. 2012, 47, 331–337
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms

J. Ruan et al.
138; therefore, they were retronecine-type PAs and/or PA
N-oxides. Furthermore, the MS2 spectra (Figure 4B) of peaks 1,
2, and 3 all exhibited characteristic ion clusters at m/z 118–120
and 136–138, suggesting that they were all retronecine-type
diester PA N-oxides. The MS2 spectra of peaks 4, 5, and 6 did
not only show the typical N-oxide ion clusters and thus were
retronecine-type PAs. These results further confirmed that the
established characteristic ion clusters of PA N-oxides can be
successfully used to identify rapidly PA N-oxides in PA-containing
natural products even without requiring the reference standards.
Furthermore, with the availability of reference standards, four
ingredients in G. pseudochina were unambiguously identified as
seneciphylline N-oxide (peak 1), senecionine N-oxide (peak 2),
seneciphylline (peak 4), and senecionine (peak 5). According to their
molecular weight and PAs found in Gynura genus,[23] peaks 3 and 6
were identified as seneciphyllinine N-oxide seneciphyllinine. The
structures of seneciphyllinine and its N-oxide are shown in Supple-
mentary Figure 3. Similarly, the herbal extract of G. japonica was also
analyzed by HPLC–MS method, and interestingly the same three PAs
and three PA N-oxides were identified (data not shown).
an official guidance or policy statement of the US Food and Drug
Administration (FDA). No official support or endorsement by the
US FDA is intended or should be inferred.
Supporting Information
Supporting information may be found in the online version of
this article.
REFERENCES
[1] P. P. Fu, H. M. Chiang, Q. Xia, T. Chen, B. H. Chen, J. J. Yin, K. C. Wen, G.
Lin, H. Yu. Quality assurance and safety of herbal dietary supple-
ments. J. Environ. Sci. Health C Environ. Carcinog. Ecotoxicol. Rev.
2009, 27, 91.
[2] P. C. Chan, Q. Xia, P. P. Fu. Ginkgo biloba leave extract: biological,
medicinal, and toxicological effects. J. Environ. Sci. Health C Environ.
Carcinog. Ecotoxicol. Rev. 2007, 25, 211.
[3] P. P. Fu, Q. Xia, G. Lin, M. W. Chou. Pyrrolizidine alkaloids--genotoxicity,
metabolism enzymes, metabolic activation, and mechanisms. Drug
Metab. Rev. 2004, 36, 1.
Previously, using HPLC–MS analysis, several PA N-oxides have
been identified by our research group[21] and others,[22,23] and
the identification of all these PA N-oxides was mainly based on
the corresponding protonated molecular ion ([M + H]+) and/or
protonated molecular dimer adduct ion ([2 M + H]+) and further
confirmed with the comparison of the reference standards. With
the mass spectra reported in these studies, we have retrospec-
tively scrutinized mass spectra of all PA N-oxides identified and
definitively identified the characteristic ion clusters at m/z
118–120 and 136–138 in the mass spectra of retrorsine N-oxide,
senecionine N-oxide, and seneciphylline N-oxide in the study of
Senecionis scandentis,[21] adonifoline N-oxide in the study of
in vitro rat microsomal metabolism of adonifoline,[22] and
seneciphyllinine N-oxide in the study of Gynura segetum.[23]
These findings further support our novel approach to use the
two characteristic mass spectrometric characteristic ion clusters
at m/z 118–120 and 136–138 for the determination of toxic
retronecine-type PA N-oxides in different samples.
[4] N. Li, Q. Xia, J. Ruan, P. P. Fu, G. Lin. Hepatotoxicity and Tumorigenicity
Induced by Metabolic Activation of Pyrrolizidine Alkaloids in Herbs.
Curr. Drug Metab. 2011, 12, 823.
[5] A. R. Mattocks. Toxicity of pyrrolizidine alkaloids. Nature 1968, 217, 723.
[6] G. Lin, Y. Y. Cui, E. M. Hawes. Characterization of rat liver microsomal
metabolites of clivorine, an hepatotoxic otonecine-type pyrrolizidine
alkaloid. Drug Metab. Dispos. 2000, 28, 1475.
[7] G. Lin, Y. Y. Cui, X. Q. Liu, Z. T. Wang. Species differences in the
in vitro metabolic activation of the hepatotoxic pyrrolizidine alkaloid
clivorine. Chem. Res. Toxicol. 2002, 15, 1421.
[8] G. Lin, Y. Y. Cui, X. Q. Liu. Gender differences in microsomal meta-
bolic activation of hepatotoxic clivorine in rat. Chem. Res. Toxicol.
2003, 16, 768.
[9] Q. Xia, M. W. Chou, G. Lin, P. P. Fu. Metabolic formation of DHP-derived
DNA adducts from a representative otonecine type pyrrolizidine alka-
loid clivorine and the extract of Ligularia hodgsonnii hook. Chem. Res.
Toxicol. 2004, 17, 702.
[10] G. Lin, J. Y. Wang, N. Li, M. Li, H. Gao, Y. Ji, F. Zhang, H. Wang, Y. Zhou, Y.
Ye, H. X. Xu, J. Zheng. Hepatic sinusoidal obstruction syndrome associ-
ated with consumption of Gynura segetum. J. Hepatol. 2011, 54, 666.
[11] M. T. Williams, B. J. Warnock, J. M. Betz, J. J. Beck, D. R. Gardner, S. T.
Lee, R. J. Molyneux, S. M. Colegate. Detection of High Levels of
Pyrrolizidine-N-oxides in the Endangered Plant Cryptantha crassipes
(Terlingua Creek Cat’s-eye) Using HPLC-ESI-MS. Phytochem. Anal.
2011, 22, 532.
[12] A. R. Mattocks. Chemistry and Toxicology of Pyrrolizidine Alkaloids.
Academic Pr: London, 1986.
[13] J. Yan, Q. Xia, M. W. Chou, P. P. Fu. Metabolic activation of retronecine
and retronecine N-oxide - formation of DHP-derived DNA adducts.
Toxicol. Ind. Health 2008, 24, 181.
[14] M. W. Chou, Y. P. Wang, J. Yan, Y. C. Yang, R. D. Beger, L. D. Williams,
D. R. Doerge, P. P. Fu. Riddelliine N-oxide is a phytochemical and
mammalian metabolite with genotoxic activity that is comparable
to the parent pyrrolizidine alkaloid riddelliine. Toxicol. Lett. 2003,
145, 239
[15] Y. P. Wang, J. Yan, P. P. Fu, M. W. Chou. Human liver microsomal reduc-
tion of pyrrolizidine alkaloid N-oxides to form the corresponding
carcinogenic parent alkaloid. Toxicol. Lett. 2005, 155, 411.
[16] R. J. Molyneux, A. E. Johnson, J. D. Olsen, D. C. Baker. Toxicity of
pyrrolizidine alkaloids from Riddell groundsel (Senecio riddellii) to
cattle. Am. J. Vet. Res. 1991, 52, 146.
[17] H. G. Mol, R. C. Van Dam, P. Zomer, P. P. Mulder. Screening of plant
toxins in food, feed and botanicals using full-scan high-resolution
(Orbitrap) mass spectrometry. Food Addit. Contam. Part A Chem.
Anal. Control Expo. Risk Assess. 2011, 28, 1405.
[18] X. Yang, L. Yang, A. Xiong, D. Li, Z. Wang. Authentication of Senecio
scandens and S. vulgaris based on the comprehensive secondary
metabolic patterns gained by UPLC-DAD/ESI-MS. J. Pharm. Biomed.
Anal. 2011, 56, 165.
CONCLUSIONS
Our study has, for the first time, found that the characteristic ion
clusters were unique determinants to distinguish between PA
N-oxides and PAs and also identify different subtypes among
retronecine-type PA N-oxides. Our findings provide a novel and
specific analytical method to differentiate PA N-oxides from PAs
in PA-containing natural products. The discrimination of PA
N-oxides from PAs will allow predict the risk of PA intoxication
and make suitable quality control of PA-containing products
and thus reducing the potential high risk of PA poisonings.
Acknowledgement
The authors greatly acknowledge the funding support from
Research Grant Council of Hong Kong Special Administrative Re-
gion, China (GRF Grant Project no. CUHK2140690) to the present
study. The financial support from the National Natural Science
Funds for Distinguished Young Scholar (No. 30925043) is also
greatly acknowledged. They thank Dr Po-Chuen Chan of NTP
for supplying riddelliine, Dr John A. Edgar of CSIRO for supplying
lasiocarpine, and Prof. Hai Shen Chen of Secondary Military
Medical University for supplying Platyphylline. This article is not
[19] C. Wang, Y. Li, J. Gao, Y. He, A. Xiong, L. Yang, X. Cheng, Y. Ma, Z.
Wang. The comparative pharmacokinetics of two pyrrolizidine
wileyonlinelibrary.com/journal/jms
Copyright © 2012 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2012, 47, 331–337

Unique ion clusters to determine PA N-oxides
alkaloids, senecionine and adonifoline, and their main metabolites
in rats after intravenous and oral administration by UPLC/ESIMS.
Anal. Bioanal. Chem. 2011, 401, 275.
chromatography/ion trap mass spectrometry. Rapid Commun.
Mass Spectrom. 2009, 23, 291.
[24] Y. Zhou, N. Li, F. F. Choi, C. F. Qiao, J. Z. Song, S. L. Li, X. Liu, Z. W. Cai,
P. P. Fu, G. Lin, H. X. Xu. A new approach for simultaneous screening
and quantification of toxic pyrrolizidine alkaloids in some
potential pyrrolizidine alkaloid-containing plants by using ultra
performance liquid chromatography-tandem quadrupole mass
spectrometry. Anal. Chim. Acta 2010, 681, 33.
[20] G. Lin, K. Y. Zhou, X. G. Zhao, Z. T. Wang, P. P. But. Determination of
hepatotoxic pyrrolizidine alkaloids by on-line high performance
liquid chromatography mass spectrometry with an electrospray
interface. Rapid Commun. Mass Spectrom. 1998, 12, 1445.
[21] S. L. Li, G. Lin, P. P. Fu, C. L. Chan, M. Li, Z. H. Jiang, Z. Z. Zhao.
Identification of five hepatotoxic pyrrolizidine alkaloids in
a
[25] C. Crews, F. Berthiller, R. Krska. Update on analytical methods for
toxic pyrrolizidine alkaloids. Anal. Bioanal. Chem. 2010, 396, 327.
[26] Y. Zhao, Q. Xia, J. J. Yin, G. Lin, P. P. Fu. Photoirradiation of
dehydropyrrolizidine alkaloids--formation of reactive oxygen species
and induction of lipid peroxidation. Toxicol. Lett. 2011, 205, 302.
[27] G. Lin, S. L. Li, M. Li, N. Li, S. Sun-Kin Chan, W. Y. Chan, Z. Z. Zhao.
Qianliguang (Senecio scandens) safety dilemma: dose is the key?
Planta Med. 2009, 75, 1107.
commonly used traditional Chinese medicinal herb, Herba
Senecionis scandentis (Qianliguang). Rapid Commun. Mass
Spectrom. 2008, 22, 591.
[22] A. Xiong, L. Yang, Y. He, F. Zhang, J. Wang, H. Han, C. Wang, S.
W. Bligh, Z. Wang. Identification of metabolites of adonifoline,
a hepatotoxic pyrrolizidine alkaloid, by liquid chromatography/
tandem and high-resolution mass spectrometry. Rapid Commun.
Mass Spectrom. 2009, 23, 3907.
[28] G. Lin, G. McKay, J. W. Hubbard, K. K. Midha. Decomposition of
clozapine N-oxide in the qualitative and quantitative analysis of
clozapine and its metabolites. J. Pharm. Sci. 1994, 83, 1412.
[23] X. Qi, B. Wu, Y. Cheng, H. Qu. Simultaneous characterization of
pyrrolizidine alkaloids and N-oxides in Gynura segetum by liquid
J. Mass. Spectrom. 2012, 47, 331–337
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms