Pyrrolizidine Alkaloids of Blue Heliotrope ... - ACS Publications

NMR analysis confirmed the identity of helioamplexine as a previously unreported indicine homologue. This is the first report of the isolation of inte...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JAFC

Cite This: J. Agric. Food Chem. 2019, 67, 7995−8006

Pyrrolizidine Alkaloids of Blue Heliotrope (Heliotropium amplexicaule) and Their Presence in Australian Honey Matheus Carpinelli de Jesus,†,⊥ Natasha L. Hungerford,‡,⊥ Steve J. Carter,§ Shalona R. Anuj,§ Joanne T. Blanchfield,† James J. De Voss,† and Mary T. Fletcher*,‡ School of Chemistry and Molecular Biosciences and ‡Queensland Alliance for Agriculture and Food Innovation (QAAFI), The University of Queensland, Brisbane, Queensland 4072, Australia § Forensic and Scientific Services, Queensland Health, Brisbane, Queensland 4108, Australia Downloaded via GUILFORD COLG on July 23, 2019 at 11:16:12 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



S Supporting Information *

ABSTRACT: Blue heliotrope (Heliotropium amplexicaule) is an invasive environmental weed that is widely naturalized in eastern Australia and has been implicated as a source of pyrrolizidine alkaloid (PA) poisoning in livestock. Less welldocumented is the potential of such carcinogenic alkaloids to contaminate honey from bees foraging on this plant species. In this study, the PA profile of H. amplexicaule plant material, determined by HRAM LC-MS/MS, revealed the presence of nine PAs and PA-N-oxides, including several PAs and PA-N-oxides of the indicine class, which have not previously been reported. The predominant alkaloid, indicine, represents 84% of the reduced PA content, with minor alkaloids identified as intermedine and the newly reported helioamplexine, constituting 7 and 9%, respectively. NMR analysis confirmed the identity of helioamplexine as a previously unreported indicine homologue. This is the first report of the isolation of intermedine, helioamplexine, and 3′-O-angelylindicine from H. amplexicaule. Also described is the identification of N-chloromethyl analogues of the major alkaloids as isolation-derived artifacts from reactions with dichloromethane. Analysis of regional-market honey samples revealed a number of honey samples with PA profiles analogous to that seen in H. amplexicaule, with measured PA contents of up to 2.0 μg of PAs per gram of honey. These results confirm the need for honey producers to be aware of H. amplexicaule as a potential PA source, most particularly in products where honey is sourced from a single location. KEYWORDS: LC-MS, pyrrolizidine alkaloid, Heliotropium amplexicaule, honey, N-chloromethyl artifacts



plantagineum L. (Paterson’s curse)7,14,15 as the primary potential source of PAs. However, in past decades, concerted eradication and biological-control programs have largely restricted the prevalence of this introduced pest species.16 In an examination of market honey, however, we have identified honey with alkaloid profiles that appeared to be consistent with a number of other PA-containing plant species also present within the Australian environment. Heliotropium amplexicaule (blue heliotrope) is one such species that has been used both intentionally and unintentionally as a floral source for Australian honey.15,17,18 This report describes the characterization of nine PAs (1−9, Figure 1) from H. amplexicaule, allowing development of a unique fingerprint to identify honey arising from this heliotrope species. These alkaloids (1−9) are all C-9 monoesters of the dehydronecine base retronecine and as such have the requisite chemistry for cytotoxicity and carcinogenicity.1 H. amplexicaule was introduced into Australia as an ornamental garden plant in the late 1800s and is now widely naturalized throughout the eastern and southeastern parts of Australia.19 Blue heliotrope is regarded as a serious environmental weed in Australia, whose spread is expanding in both

INTRODUCTION Pyrrolizidine alkaloids (PAs) encompass a class of more than 600 different alkaloids found in flowering plants of the families Boraginaceae, Fabaceae (tribe Crotalarieae), and Asteraceae (tribe Senecioneae).1 These alkaloids are potent carcinogens and can not only poison livestock grazing on such plants2 but also enter the human food chain either directly through herbal medicines or coharvesting of PA-containing plants with edible grains or, to a lesser extent, indirectly through commodities such as meat, milk, eggs, and honey.1,3,4 Indeed, the potential PA contamination of honey from bees foraging on plants containing these potent carcinogens has elicited considerable attention in recent years, with several intensive market surveys conducted5−7 and various regulations now applied worldwide.3,4,8,9 When bees collect nectar and pollen from PAcontaining plants, these alkaloids can be transferred into honey, with the relative contributions from pollen and nectar not always clear.9 Nectar from Echium vulgare has, for example, recently been demonstrated to be the major (but not exclusive) source of PAs in honey from this floral source.10 Interestingly, however, in plants PAs occur mainly in the form of the corresponding N-oxides9,11 but are found in honey as the more toxic free alkaloids; this reduction of the N-oxide has been demonstrated to occur in the digestive tract of honeybees12 and during storage.13 Previous studies examining the presence of pyrrolizidine alkaloids in Australian honey have focused on Echium © 2019 American Chemical Society

Received: Revised: Accepted: Published: 7995

April 4, 2019 May 24, 2019 May 30, 2019 May 30, 2019 DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

cultivated pastures and crop fields. A targeted biocontrol program utilizing insect species from the native host range in South America has not been successful in reducing this spread.20 This heliotrope is poisonous to livestock, and production losses occur because of declines in animal performance (and death) as a result of PA hepatotoxicity from infested pastures or contaminated stock feed.21,22 All livestock can be affected, including, in order of susceptibility, horses, pigs, cattle, sheep, and goats.22 Previous studies have reported the presence of indicine (3, Figure 1) as the major alkaloid in this species, together with occasional mention of a minor alkaloid, which has been either reported as “unknown”21 or purported to be heliotrine (10).15 The purpose of this study was to determine the full range of PAs present in this plant so as to provide a unique alkaloid LCMS/MS profile that could be used for comparison with market honey samples. Identification of novel minor-alkaloid components present in both the honey and the plant source enables profile matching to provide greater assurance of alkaloid plant source. Identification of these sources is crucial in order to advise the honey industry and provide direction with regard to better siting of hives to avoid potential contamination issues. Livestock and domestic animals may be exposed to PAs through the consumption of forage and roughage contami-

Figure 1. Compounds 1−9 from H. amplexicaule, compared with the previously purported component, heliotrine (10).

Table 1. Details of Pyrrolizidine Alkaloids Used in the Orbitrap LC-MS/MS Analysis of H. amplexicaule and Honey, Including Formulae, Retention Times, Precursor Ions Used for Quantitation, and Confirmatory Product Ions compound echimidine echimidine N-oxide erucifoline erucifoline N-oxide europine europine N-oxide helioamplexine helioamplexine N-oxide heliotrine heliotrine N-oxide indicine indicine N-oxide and intermedine N-oxidea intermedine jacobine jacobine N-oxide lasiocarpine lasiocarpine N-oxide lycopsamine lycopsamine N-oxide monocrotaline monocrotaline N-oxide retrorsine retrorsine N-oxide senecionine senecionine N-oxide seneciphylline seneciphylline N-oxide senecivernine senecivernine N-oxide senkirkine trichodesmine

formula

average RT (min)

precursor ion (MH+)

C20H31NO7 C20H31NO8 C18H23NO6 C18H23NO7 C16H27NO6 C16H27NO7 C16H27NO5 C16H27NO6 C16H27NO5 C16H27NO6 C15H25NO5 C15H25NO6

12.86 12.80 6.23 8.03 6.97 7.86 9.21 10.42 9.56 10.46 6.67 8.20

398.2173 414.2122 350.1598 366.1547 330.1911 346.1860 314.1962 330.1911 314.1962 330.1911 300.1806 316.1755

120.0809 396.2004 322.1642 278.1386 254.1385 328.1743 156.1017 172.0966 156.1017 172.0966 156.1019 226.1437

83.0497 352.1745 220.1329 218.1172 156.1019 270.1328 138.0913 155.0938 138.0913 155.0938 138.0914 172.0968

55.0550 254.1379 164.1066 164.1067 138.0914 256.1172 120.0808 138.0913 120.0808 138.0913 120.0810 155.0941

220.1326 138.0911 136.0756 120.0810 172.0964 94.0655 111.0913 94.0655 111.0913 94.0656 138.0914

137.0833 120.0807 119.0729 96.0812 155.0937

94.0653 82.0657 111.0682

94.0656

C15H25NO5 C18H25NO6 C18H25NO7 C21H33NO7 C21H33NO8 C15H25NO5 C15H25NO6 C16H23NO6 C16H23NO7 C18H25NO6 C18H25NO7 C18H25NO5 C18H25NO6 C18H23NO5 C18H23NO6 C18H25NO5 C18H25NO6 C19H27NO6 C18H27NO6

6.26 6.56 7.91 14.92 16.14 6.80 8.65 2.88 7.19 9.13 9.64 11.13 11.83 9.52 10.46 10.84 11.45 13.62 8.79

300.1806 352.1755 368.1704 412.2330 428.2279 300.1806 316.1755 326.1598 342.1547 352.1755 368.1704 336.1806 352.1755 334.1649 350.1598 336.1806 352.1755 366.1911 354.1911

210.1488 308.1485 296.1485 238.1435 410.2168 156.1017 172.0964 280.1548 314.1590 324.1802 220.1340 308.1864 324.1825 306.1706 322.1656 308.1848 324.1795 168.1020 308.1857

156.1019 280.1539 190.1222 120.0810 352.1746 138.0914 155.0937 237.1354 296.1487 138.0913 154.0862 120.0809 220.1332 120.0811 246.1495 153.0907 220.1327 150.0915 223.1203

138.0914 262.1432 139.0989 138.0914 328.1753 120.0808 138.0911

120.0810 234.1483 121.0885 156.1020 254.1384 94.0655 136.0755

94.0656 155.1063 120.0807 94.0656 220.1333

137.0835

236.1274 120.0808

137.0833 94.0655

119.0729

138.0911 154.0859

120.0807 136.0755

94.0654 120.0807

222.1489

164.1071

121.0889

product ions

94.0653

94.0654

a

Not resolved. 7996

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry Table 2. High-Resolution Accurate-Mass (HRAM) Data for Pyrrolizidine Alkaloids in H. amplexicaule typical RT (min)

molecular-ion formula

calculated MH+

observed MH+

5′-hydroxyindicine (1)

2.22

C15H25NO6 + H+

316.1755

316.1750

intermedine (2)

6.26

C15H25NO5 + H+

300.1805

300.1808

indicine (3)

6.67

C15H25NO5 + H+

300.1805

300.1805

indicine N-oxide (4)

8.20

C15H25NO6 + H+

316.1755

316.1752

helioamplexine (5)

9.21

C16H27NO5 + H+

314.1962

314.1959

helioamplexine N-oxide (6)

10.42

C16H27NO6 + H+

330.1911

330.1908

3′-O-angelylindicine (7)

14.68

C20H31NO6 + H+

382.2224

382.2223

3′-O-angelylindicine N-oxide (8) 3′-O-angelylhelioamplexine (9)

15.55

C20H31NO7 + H+

398.2173

398.2166

14.71

C21H33NO6 + H+

396.2381

396.2372

alkaloid

Thermo Fisher Scientific. Analyte detection was performed by positive electrospray ionization (ESI) with a spray voltage of 3500 V and a vaporizer temperature of 400 °C. MS analysis was run with arbitrary pressures for sheath gas (48), auxiliary gas (11), and sweep gas (2); a spray voltage of 3.5 kV; a capillary temperature of 320 °C; and an auxiliary-gas heater at 350 °C. Full-scan−dd-MS2 mode was used. Full scans were conducted at a resolution of 70 000 fwhm (at m/z 200), with an AGC target of 1.00 × 106. The maximum time for accumulating ions per scan event was 10 ms, with a scan range of m/z 75−1125. Data-dependent acquisition (dd-MS2) was conducted at a resolution of 17 500 with an AGC target of 1.00 × 106. The maximum time for accumulating ions per scan event was 50 ms. The normalized collision energy (nce) was set to 50%, and an isolation window of m/z 1.0 was utilized. Dynamic exclusion was set to 3 s, preventing subsequent triggering of the same ion in data-dependent scans. A maximum of the five most abundant precursors could be selected for dd-MS2 per scan event. Pyrrolizidine alkaloid levels in honey and plant material were determined by quantification against certified PA standards. Calibration curves were obtained for each of the 30 pyrrolizidine alkaloid standards injected at 5, 10, 20, 50, 100, and 200 ppb (in duplicate or triplicate). Quantitation involved comparison of the precursor-parent-ion intensities to those in the standard curves, with squared correlation coefficients (R2) in the range of 0.9932−0.9997. Plant and honey extracts were analyzed to detect pyrrolizidine alkaloids and their N-oxides by matching of retention times with those of the corresponding standards; they were identified by their precursor parent ions (M + H+) and confirmed by the presence of product ions (Table 1). The identities of additional alkaloids could be confirmed further by use of the high-resolution accurate-mass data provided by the Q Exactive mass spectrometer, enabling the determination of the elemental compositions of parent and product ions. Plant Material. H. amplexicaule was collected from Aspley, suburban Brisbane. Taxonomical identification was confirmed by the Queensland herbarium, with a voucher specimen (AQ522457) incorporated into their permanent collection. The sample of H. amplexicaule collected for analysis (3 kg) was a composite of stems, leaves, and flowers, which was freeze-dried, milled, and stored frozen prior to analysis. Plant-Alkaloid Extraction. Milled plant material (60 g) was extracted in an ultrasonic bath for 1 h with methanol (2 × 400 mL). The combined methanol extracts were concentrated, and the residue was suspended in 3.2% HCl (100 mL) and washed with diethyl ether (2 × 100 mL) and dichloromethane (2 × 100 mL). The aqueous

nated with PA-containing plants or seeds, but there is limited carry over from animal feed into meat, milk, and eggs.3 Honey, however, constitutes a significant source of human PA exposure,4 and identification of the source plant is of critical importance in minimizing this exposure.



mass-spectral data (relative intensity, %) 316.1750 (1), 156.1018 (29), 139.0991 (9), 138.0913 (32), 120.0809 (18), 108.0811 (1), 96.0811 (2), 96.0448 (1), 95.0734 (4), 94.0655 (100), 82.0657 (4) 300.1808 (5), 156.1020 (51), 139.0992 (11), 138.0914 (29), 120.0810 (17), 112.0760 (2), 96.0813 (4), 95.0734 (4), 94.0656 (100), 82.0657 (5) 300.1805 (5), 156.1019 (51), 139.0992 (11), 138.0914 (27), 120.0809 (19), 112.0759 (2), 96.0812 (4), 95.0734 (4), 94.0656 (100), 82.0657 (5) 316.1752 (33), 172.0967 (100), 155.0940 (17), 154.0862 (8), 138.0913 (54), 137.0836 (6), 136.0757 (15), 111.0681 (12), 94.0655 (21), 93.0577 (15) 314.1959 (3), 156.1019 (51), 139.0992 (12), 138.0914 (23), 120.0809 (17), 112.0759 (3), 96.0812 (4), 95.0734 (6), 94.0656 (100), 82.0657 (6) 330.1908 (23), 172.0967 (100), 155.0940 (20), 154.0862 (7), 138.0914 (46), 136.0757 (14), 111.0681 (15), 94.0655 (20), 93.0577 (13), 82.0419 (6) 282.1697 (1), 156.1019 (12), 139.0992 (4), 138.0914 (18), 120.0809 (20), 96.0448 (2), 95.0734 (2), 94.0656 (100), 83.0497 (6), 82.0657 (3), 55.0550 (4) 298.1647 (22), 172.0967 (100), 155.0940 (18), 154.0862 (10), 138.0913 (79), 136.0757 (34), 111.0681 (23), 94.0656 (78), 93.0577 (26), 83.0497 (15) 296.1850 (1), 156.1015 (11), 139.0987 (4), 138.0910 (20), 120.0806 (20), 97.0650 (6), 95.0731 (2), 94.0653 (100), 83.0494 (5), 82.0654 (3), 55.0549 (3)

MATERIALS AND METHODS

Chemicals and Solvents. A total of 30 pyrrolizidine alkaloid standards were utilized in a high-resolution accurate-mass (HRAM) LC-MS/MS screen. Echimidine, erucifoline, europine, heliotrine, indicine, intermedine, jacobine, lasiocarpine, lycopsamine, monocrotaline, retrorsine, senecionine, seneciphylline, senecivernine, and their respective N-oxides were purchased together with senkirkine and trichodesmine from Phytolab GmbH & Company KG (Vestenbergsgreuth, Germany) and had purities >89%. All other chemicals and solvents were of analytical-reagent or HPLC-grade purity. The water used for sample preparation and HPLC was Milli-Q purified (Millipore). Honey Samples. Honey samples were purchased between July 2016 and May 2017 from Queensland supermarkets, fruit shops, local markets, and producers. Honey Extraction. Honey samples (1 g) were dissolved in H2SO4 (0.05 M, 10 mL) and centrifuged, and the supernatants were applied to Agilent SPE Bond Elut 100 mg LRC-SCX columns. The SPE cartridges were washed with water (10 mL) and methanol (10 mL), and pyrrolizidine alkaloids were then eluted with 3% ammonia in methanol (3 mL). The eluate was evaporated to dryness under nitrogen, and the residue was reconstituted in 5% methanol in water (1 mL) for HRAM LC-MS/MS analysis. HRAM LC-MS/MS Analysis. Samples were analyzed on a Thermo Scientific Vanquish UHPLC combined with a Q Exactive Orbitrap high-resolution accurate-mass (HRAM) spectrometry system. LCMS/MS separation was carried out on a Kinetex XB-C18 analytical column (100 × 2.1 mm, 2.6 μm, 100 Å) at 5 °C. The analysis conditions were as follows: A binary solvent system was used, which included solvent A (5 mM aqueous ammonium formate and 0.1% formic acid) and solvent B (95%, v/v, methanol/water with 5 mM ammonium formate and 0.1% formic acid). Compounds were eluted from the column at 0.3 mL min−1 with mobile phase B held at 5% for 3 min, followed by linear gradients of B from 5 to 50% (3−15 min), 50−80% (15−18.5 min), and 80−100% (18.5−19 min); it was held at 100% for 30 s and then reduced to 5% over 6 s, where it was held until elution was stopped at 23.5 min. The instrument control and data acquisition and analysis were conducted using Tracefinder 4.1 from 7997

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry Table 3. High-Resolution Accurate-Mass (HRAM) Data for H. amplexicaule Pyrrolizidine Alkaloid Artifacts alkaloid artifact N-chloromethyl indicine (11)

N-chloromethyl helioamplexine (12) N-chloromethyl 3′O-angelylindicine (13)

typical RT (min)

molecular-ion formula

calculated M+

observed M+

7.28

C16H2735ClNO5+

348.1572

348.1562

C16H2737ClNO5+

350.1543

350.1533

9.28

C17H2935ClNO5+

362.1729

362.1717

13.84

C17H2937ClNO5+ C21H3335ClNO6+

364.1699 430.1991

364.1688 430.1978

C21H3337ClNO6+

432.1961

432.1951

mass-spectral data (relative intensity, %) 348.1562 (19), 312.1792 (0.3), 204.0781 (100), 187.0754 (12), 186.0676 (30), 168.1013 (11), 152.1065 (11), 150.0909 (12), 144.0571 (33), 143.0493 (22), 108.0808 (27) 350.1533 (21), 206.0749 (100), 189.0724 (13), 188.0645 (31), 168.1014 (12), 152.1065 (11), 150.0909 (13), 146.0541 (32), 145.0463 (22), 108.0807 (28) 362.1717 (13), 204.0781 (100), 187.0754 (14), 186.0676 (25), 168.1015 (11), 152.1066 (13), 150.0910 (11), 144.0571 (28), 143.0493 (28), 108.0808 (26) MS2 not observed 430.1978 (3), 330.1452 (5), 204.0780 (100), 186.0676 (28), 168.1015 (16), 150.0910 (22), 144.0571 (41), 143.0493 (13), 142.0415 (12), 108.0808 (42), 83.0494 (27), 55.0549 (16) MS2 not observed

Table 4. 1H NMR Data for Alkaloids 3−5 and 7 (500 MHz, CDCl3) position

3

3H+

4

5

7

2 3a 3b 5a 5b 6a 6b 7 8 9a 9b 3′ 4′ 5′ 6′a 6′b 7′ 8′ 3″ 4″ 5″

5.90 (br s) 3.93 (d, 15.5) 3.30−3.24 (m) 3.42 (dd, 5.89, 1.40) 2.78−2.69 (m)

5.86 (s) 4.34 (d, 15.5) 3.58 (d, 15.5) 3.75 (t, 8.3) 3.02 (td, 11.3, 6.2) 2.20−2.15 (m) 2.12−2.07 (m) 4.54 (td, 3.7, 1.5) 4.85 (s) 5.21 (ABd, 12.9) 4.58 (ABd, 12.9) 4.13 (q, 6.4) 1.20 (d, 6.4) 2.07−2.04 (m)

5.93 (s) 4.74 (d, 15.0) 4.51 (d, 15.0) 4.13−4.07 (m) 3.82 (td, 12.0, 5.65) 2.76−2.69 (m) 2.19−2.13 (m) 4.77 (br s) 5.10 (s) 5.30 (ABd, 13.1) 4.66 (ABd, 13.1) 4.17 (q, 6.3) 1.25 (d, 6.3) 2.00 (sept, 7.0)

5.77 (d, 1.5) 4.20−4.16 (m) 3.45 (d, 6.1) 3.62 (t, 9.5) 2.94−2.89 (m) 2.17 (d, 3.4) 1.57−1.45 (m) 4.47 (br s) 4.54 (br s) 4.79 (ABd, 13.7) 4.67 (ABd, 13.7) 5.34 (q, 6.4) 1.31 (d, 6.4) 2.07−2.05 (m)

0.95 (d, 7.1)

0.94 (d, 7.1)

0.96 (d, 7.0)

0.91 (d, 7.1)

0.93 (d, 7.1)

0.91 (d, 7.0)

5.88 (s) 4.26 (d, 15.5) 3.54 (ddd, 15.5, 4.3, 1.8) 3.66 (ddd, 10.0, 7.7, 1.5) 2.97 (ddd, 12.2, 10.0, 6.1) 2.20−2.13 (m) 2.12−2.07 (m) 4.49 (td, 3.7, 1.5) 4.72 (s) 5.20 (ABd, 12.8) 4.55 (ABd, 12.8) 4.16 (q, 6.4) 1.19 (d, 6.4) 1.81−1.76 (m) 1.38−1.32 (m) 1.25−1.19 (m)a 0.91 (t, 6.8) 0.94 (d, 6.8)

2.04−1.92 (m) 4.28 4.16 5.09 4.57 4.03 1.17 2.13

(br s) (br s) (d 12.7) (d 12.7) (q, 6.4) (d, 6.4) (hept, 7.1)

1.02 (d, 7.0) 0.94 (d, 7.0) 6.09 (qq, 7.4, 1.5) 1.96 (dq, 7.4, 1.5) 1.82 (quintet, 1.5)

a

Splitting obscured by overlapping signals.

layer was divided into two equal portions (100 mL). The first portion was made alkaline with aqueous ammonia (2.8%, 100 mL) and extracted with DCM (2 × 100 mL). The second portion was treated with zinc dust (0.3 g) in 3.2% HCl (100 mL) and stirred at room temperature overnight (16 h). This reduced extract was made alkaline with aqueous ammonia (2.8%, 150 mL) and extracted with DCM (2 × 100 mL). The organic extracts from each portion were concentrated separately under a stream of N2, and the alkaloid content in each extract was determined by HRAM LC-MS/MS analysis. Preparative HPLC Separation. DCM extracts were chromatographed at room temperature on a Shimadzu LC-10AT series system equipped with a semipreparative column (Phenomenex, C-18 (2), 250 × 10.00 mm, 5 μm, 100 Å) and with Shimadzu/ELSD-LT and Shimadzu/SPD-20A Prominence UV detectors. The analysis conditions were as follows: A binary solvent system of solvent A (15 mM aqueous ammonium acetate) and solvent B (acetonitrile) was used. Compounds were eluted from the column at 1.5 mL min−1 with mobile phase B held at 5% for 5 min; this was followed by a linear gradient of B from 5 to 50% (5−45 min), where it was held for 2 min, before being increased from 50 to 70% over 2 min; the concentration of B was held at 70% then until elution was stopped at 59.5 min. The UV detector monitored wavelengths 210 and 254 nm.

The ELSD was operated at 50 °C and 2.5 bar nitrogen, with 10× gain. A fraction collector was used to collect 16 individual peaks or fractions, eluting between 32−53 min, across peaks demonstrated by LC-MS to contain one or more alkaloids. LC-MS analysis of collected fractions was carried out on a Shimadzu 2020 series LC-MS system equipped with a Shimadzu Diode Array detector, an SPD-20A Prominence UV−vis detector (Shimadzu, W lamp), and an analytical column (Phenomenex Polar C18, 100 × 2.1 mm, 1.6 μm, 100 Å) maintained at 45 °C. The analysis conditions were as follows: A binary solvent system of mobile phase A (water and 0.1% formic acid) and solvent B (acetonitrile and 0.1% formic acid) was used. Compounds were eluted from the column at 0.25 mL min−1 with mobile phase B held at 5% for 5 min; this was followed by a linear gradient of 5−35% B (5−20 min), after which the percentage of B was held constant for 2 min. Selected-ionmonitoring (SIM) mode was activated to monitor positive ions. The UV detector monitored 213 and 280 nm. Alkaloid Identification. All compounds separated from the reduced H. amplexicaule extract were subjected to HRAM LC-MS/ MS comparisons with standard alkaloids (Table 1). For unknown alkaloids, accurate-mass and MS interpretation was used to identify compounds (Tables 2 and 3). Further confirmation was obtained via 1D- and 2D-nuclear-magnetic-resonance (NMR) analysis performed 7998

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

Table 7. 13C NMR Data for N-Chloromethyl Artifacts 11 and 12 (175 MHz, CDCl3)

on a 500 MHz Bruker Avance system using 5 mm BBFO probe or a Bruker Avance DRX 700 MHz instrument with a 5 mmTXI Zgard probe; samples were prepared in CDCl3 (Tables 4−7).

Table 5. 13C NMR Data for Alkaloids 3, 5, and 7 (125 MHz, CDCl3) position

3

5

7

1 2 3 5 6 7 8 9 1′ 2′ 3′ 4′ 5′ 6′ 7′ 8′ 1″ 2″ 3″ 4″ 5″

132.67 130.15 62.81 53.7 36.14 71.18 78.54 62.75 175.57 82.74 69.16 16.55 32.32 17.1 17.42

132.48 128.06 61.98 53.14 35.98 70.86 78.02 61.98 175.19 83.05 69.70 17.45 39.80 23.90 12.58 12.79

132.29 127.22 61.40 53.47 33.73 70.70 77.72 62.05 174.95 82.09 72.42 14.06 36.03 17.28 17.05

Table 6. 1H NMR Data for N-Chloromethyl Artifacts 11 and 12 (700 MHz, CDCl3) 11

12

2 3a 3b 5a 5b 6a 6b 7 8 9a 9b 10a 10b 3′ 4′ 5′ 6′a 6′b 7′ 8′

5.82 (s) 4.72 (d, 15.5) 4.20−4.17 (m) 4.15 (dd, 11.3, 6.2) 3.73 (td, 11.3, 6.2) 2.61 (dd, 13.7, 11.3, 6.2, 2.7) 2.40−2.29 (m, 13.7, 2.7) 4.96 (d, 3.7) 5.57 (s) 5.08 (ABd, 13.0) 5.02 (ABd, 13.0) 5.36 (d, 10.3) 5.03 (m) 4.13 (q, 6.5) 1.21 (d, 6.5) 1.83 (sept, 7.0)

5.85 (s) 4.72 (d, 15.8) 4.23−4.19 (m) 4.18−4.15 (m) 3.75 (td, 11.3, 6.0) 2.86−2.76 (m) 2.44−2.59 (m) 5.07 (d, 9.8) 5.70 (s) 5.16 (ABd, 13.0) 4.97 (ABd, 13.0) 5.55 (d, 10.0) 5.07 (d, 10.0) 4.18 (q, 6.4) 1.22 (d, 6.4) 1.71−1.54 (m) 1.61−1.44 (m) 1.29−1.24 (m) 0.88 (t, 7.4) 0.90 (d, 6.5)

1.00 (d, 7.0) 0.92 (d, 7.0)

11

12

1 2 3 5 6 7 8 9 10 1′ 2′ 3′ 4′ 5′ 6′ 7′ 8′

136.03 121.68 69.56 61.79 34.62 70.13 89.19 60.33 68.47 175.04 85.00 71.40 15.61 33.43 17.96 17.49

134.54 121.99 69.45 62.02 34.21 70.57 88.71 60.47 68.61 175.41 84.91 71.41 16.64 40.66 23.41 12.33 13.44

portion underwent reduction with zinc to provide free base alkaloids, whereas the remainder yielded an N-oxide-rich fraction. The free base PAs were purified by HPLC, coupled with an evaporative light-scattering detector (ELSD), as the PAs lack a significant chromophore. This yielded 10 PA-containing fractions for structural characterization, representing a combined 2% of the plants’ dry weight. HRAM LC-MS/MS and NMR evidence enabled the identification of alkaloids 1−9 (Figure 1). These (1−9) included six pyrrolizidine alkaloids, with the corresponding N-oxides 4, 6, and 8 of the three major alkaloids, 3, 5, and 7, also being detected as minor components in the reduced extract. Indicine (3).21,24−26 HRAM LC-MS/MS analysis revealed the major alkaloid in H. amplexicaule matched the molecular weight of indicine C15H25NO5 + H+, MH+ (found: 300.1805, calculated for C15H25NO5 + H+: 300.1805). (All subsequent calculated−found comparisons are given in Tables 2 and 3.) This compound was the most abundant in the LC-MS analysis of the free base forms (Figure 2), and it coeluted with authentic indicine (RT = 6.67 min). The HRMS for indicine (3) displayed m/z 94.0656 (base peak), 120.0809, 138.0914, and 156.1019, which are characteristic of C-9 monoesters of retronecine.27,28 The 1H and 13C NMR analysis (Tables 426,29 and 524) agreed with the literature, confirming both the retronecine core and the trachelanthic acid esterification at C-9.25 The 1H NMR for protonated indicine (3H+) is also reported in Table 4, with protons alpha to the nitrogen (H-3, H-5, and H-8) displaying downfield shifts relative to 3. H-2, H7, and H-8 were all observed as broad singlets. However, in the COSY, cross-couplings were observed between H-8 and H-7, between H-8 and the higher-field H-3 (homoallylic coupling), and between H-8 and H-2 (allylic coupling). H-7 also displayed cross-coupling with the lower-field H-6. Indicine N-Oxide (4). Because of the high abundance of the N-oxide of indicine in the plant material, small amounts of indicine N-oxide (4) were left unreduced by the zinc and were thus detected in the reduced fraction (Figure 2). Indicine Noxide (4) was confirmed by HRAM LC-MS/MS comparison with an authentic standard (Tables 1 and 2) and 1H NMR comparison with the literature (Table 4).25 As observed for

167.06 127.36 139.78 15.82 20.51

position

position



RESULTS AND DISCUSSION Alkaloid Composition of H. amplexicaule. The milled, aerial parts of blue heliotrope (H. amplexicaule) collected from a suburban area of Brisbane, Australia, were extracted with methanol. After evaporation, the crude extract was resuspended in acid and washed with organic solvents to produce a PA-rich aqueous partition. Because PAs occur naturally in the N-oxide form in most plants,2,23 the extract was divided and a 7999

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

Figure 2. HRAM LC-MS/MS chromatograms of the pyrrolizidine alkaloid profile of H. amplexicaule (A) without zinc reduction and (B) with zinc reduction.

fragment; however, comparison of the 1H and 13C NMR data for alkaloid 5 confirmed that, though it was similar, it was not identical to heliospathine.30 The 1H NMR spectrum of 5 (in CDCl3) displayed characteristic peaks for the retronecine core at δH 5.88 (br s, H-2), 4.72 (br s, H-8), and 4.49 (td, H-7). COSY interactions were observed between H-8 and H-7 and between H-2 and H-8, with the latter resulting from an allylic interaction; the resolution did not allow for the J coupling to be measured. The carbons associated with these protons were observed via 1D 13C NMR and assigned with the aid of HSQC to δC 128.06 (C-2), 70.86 (C-7), and 78.02 (C-8). These protons then allowed for the assignment of the remaining protons in the bicyclic core. The diastereomeric hydrogens of the C-6 methylene were seen at δH 2.20−2.13 (m, H-6a) and 2.12−2.07 (m, H-6b), identified as a result of a COSY interaction with H-7. HSQC confirmed that both protons were connected to the same carbon at δC 35.98 (C-6). Further inspection of H-6 COSY correlations then allowed assignment of δH 3.66 (ddd, H-5a) and 2.97 (ddd, H-5b).31 The signals for the remaining protons in the pyrrolizidine core were assigned

protonated indicine (3H+), protons alpha to the nitrogen (H-3, H-5, and H-8) in 4 displayed downfield shifts relative to 3 (Table 4). Indicine N-oxide (4) was also the major component in the nonreduced plant extract (Figure 2A). Indicine N-oxide (4) displayed m/z 316.1752 (C15H25NO6 + H+, MH+) and a base peak of m/z 172.0967 typical of a C-9 monoester of retronecine N-oxide.27 Helioamplexine (5). This previously unknown compound was the second most abundant compound in the reduced plant extract (Figure 2). HRAM LC-MS/MS analysis of helioamplexine (5) yielded a positive-ion at m/z 314.1959, corresponding to the molecular ion C 16H 27NO 5 + H+ (MH+), with a retention time (RT = 9.21 min) clearly different from that of the authentic standard of isomeric heliotrine (10, RT = 9.56 min). The MS/MS fragmentation of the molecular ion of 5 yielded a positive ion at m/z 156.1019 (C8H14NO2+) characteristic of a retronecine base and resulting from the loss of 158.0940 Da (C8H14O3). Such a fragment loss is observed in the related alkaloid heliospathine,30 which loses curassavic acid, a 2-(1′-hydroxyethyl)-2-hydroxy-3-methylpentanoic acid 8000

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

identified on the basis of the HRAM LC-MS/MS MH+ (found: 330.1908, calculated C16H27NO6 + H+: 330.1911) and by analogy with the indicine (3)−indicine N-oxide (4) pair already identified. Alkaloid 6 exhibited the mass fragmentation pattern characteristic of an N-oxide (Tables 1 and 2), with a typical base peak of 172.0967 (C8H14NO3+) characteristic of C-9 monoesters of retronecine N-oxides.27,28 Intermedine (2). Separation of minor alkaloids from the plant extract provided an impure mixture of known alkaloid 2, which was analyzed by HRMS and NMR. HRAM LC-MS/MS showed molecular ion m/z 300.1808 (C15H25NO5 + H+, MH+; Table 2) eluting at 6.26 min, which was assigned as intermedine (2). Alkaloid 2 coeluted with the authentic intermedine standard with identical HRAM LC-MS/MS fragmentation (Tables 1 and 2), confirming the structural assignment. The fragment loss of 144.0788 Da (C 7 H 12 O 3 ) in intermedine (2) is equivalent to the α-cleavage of trachelanthic acid, yielding the retronecine core molecular ion m/z 156.1020 (C8H12NO2+). This fragmentation pattern is identical to that of indicine and is consistent with the diastereomeric relationship between intermedine and indicine, with intermedine having a (2′S, 3′R) configuration for trachelanthic acid and indicine (3) having (2′R, 3′S) stereochemistry (Figure 1). 3′-O-Angelylindicine (7). This isolated compound was analyzed with HRAM (RT = 14.68 min), and its molecular formula of C20H31NO6 was identified from the molecular ion m/z 382.2224 (MH+). The molecular formula suggested a retronecine-like core bearing two necic acids, potentially suggesting a C-7, C-9 diester. The molecular formula was consistent with the presence of an angelic or tiglic acid moiety in addition to the trachelanthic acid ester seen in indicine (3), the major PA present in H. amplexicaule. However, the HRMS for 7 displayed m/z 94.0656 (base peak), 120.0809, 138.0914 and 156.1019 fragments, which are characteristic of C-9 monoesters of retronecine.27,28 By contrast, 7-O-angelyl diester echiumine and isomers28 display prominent m/z 120 base peaks. This m/z 120 base peak was also displayed by our LCMS/MS method for the 7-O-angelyl diester standards echimidine and lasiocarpine, with very small peaks at 156.1018 and 138.0915. Such 7-O-angelyl ester mass spectra also feature 7-O-angelyl-containing fragments at m/z 220.1332 and 238.1436.32 In alkaloid 7, a small peak at 282.1697 was observed as a result of C-9 intranecyl loss of an angelic acid or tiglic acid moiety (100.0527 Da) with a further loss of 126.0678 Da yielding the core retronecine fragment of m/z 156.1019. This raised the possibility that 7 arose from angelic or tiglic acid esterification within the necic acid moiety of a C-9 monoester PA. In the 1H NMR (Table 4) of alkaloid 7, H-2, H-7, and H-8 all were observed as broad singlets, with the long-range couplings between H-8 and H-2 and between H-8 and H-3 observable in the COSY, enabling confirmation of the assignments of these signals. The 1H NMR spectrum of the rest of the necine base was similar to those of the other PAs in H. amplexicaule (Table 4), although the chemical-shift similarities between 7 and 3H+ for H-3, H-5, and H-8 imply that the NMR of 7 corresponds to the protonated form of the molecule. The tabulated NMR assignments (Tables 4 and 5) confirmed that 7 was a retronecine derivative. The NMR data also suggested that a trachelanthic acid moiety was attached to C-9, as seen in indicine (3). H-3′ was present as a quartet at 5.34 ppm, and the C-4′ methyl was

to H-3a (δH 4.26) and H-3b (δH 3.54), as their shifts were consistent with allylic methylene protons vicinal to the bridging nitrogen. This assignment was confirmed by the HSQC correlation of these two proton signals to the same carbon signal, δC 61.98 (C-3). Methylene C-9, providing the ester link between the necine base and the necic acid, was observed as an AB doublet at δH 5.20 (d, J = 12.8 Hz, H-9a) and 4.55 (d, J = 12.8 Hz, H-9b). These proton and carbon assignments were in close accordance with those observed for the retronecine core of indicine (Table 4). Thus, the key structural difference between alkaloid 5 and indicine (3) lay in the presence of an extra carbon in the necic acid. The necic acid’s methyl doublets at δH 1.19 (d, H-4′) and 0.94 (d, H-8′) were correlated via HSQC to δC 17.45 (C-4′) and 12.79 (C8′), respectively. The secondary OH group was located at C-3′ (δC 69.70), with an expected downfield chemical shift, and associated with the proton at δH 4.16 (d, H-3′). The remaining methine signal for 5 was observed at δH 1.81−1.76 (m, H-5′) and displayed correlations in the COSY spectrum (H6′a, H6′b) and at HMBC δC 39.80 (H-4′, H-6′a, H-6′b, H-8′). The protons of the diastereotopic methylene C-6′ at δH 1.38−1.32 (m, H-6′a) and 1.25−1.19 (m H-6′b) were characterized as a result of their COSY (H-6′a−H-6′b, H-6′ab−H-7′, H-6′ab− H-5′) and HSQC correlations (δC 23.90: H-6′ab−C-6′). These coupled to an apparent triplet at δH 0.91 (t, H-7′) corresponding to the neighboring methyl protons. Correlations observed in the HMBC spectrum of 5 between δC 175.19 (C1′) and δH 5.20 (d, H-9a), δH 4.55 (d, H-9b), δH 4.16 (q, H3′), and δH 1.78 (ddt, H-5′) established the carbonyl connection between the necine base and the necic acid. HMBC correlations with H-4′ and H-8′ also allowed C-2′ at δC 83.05 (H-4′ and H-8′) to be identified. Because of the similarities in proton chemical shift, connectivity, and molecular mass, compound 5 was likely to be either heliospathine30 or a closely related stereoisomer. Interrogation of the reported chemical shifts for protons and carbons of heliospathine, indicine (3), and intermedine (2) against 5 led to the stereochemical assignment of the necic acid substituent as C-2′R and C-3′S in 5, compared with the C-2′S, C-3′S configuration reported for heliospathine.30 The structure of 5 was therefore elucidated as the C-6′ homoanalogue of indicine with the same stereochemistry at C-2′ and C-3′ (Figure 1) and thereby named helioamplexine after the plant in which it was found. In helioamplexine (5), the necic acid is therefore a previously unreported 6′-homotrachelanthic acid. Comparison with authentic heliotrine (10) confirmed that this alkaloid, with the same molecular ion of C16H27NO5 + H+ (MH+), was not detected in the H. amplexicaule extracts, and it is presumed that the previous reports15 of heliotrine (10) in this plant were misidentification of helioamplexine (5). In 5, as in proton indicine (3H+), H-2, H-7, and H-8 were observed as broad singlets, with H-8 displaying diagnostic 4J coupling to H-2 and 5J coupling to one H-3, as observed in the COSY spectrum. The similarity in chemical shifts in Table 4 between 5 and 3H+ for H-3, H-5, and H-8 suggests that the NMR of helioamplexine (5) is for the protonated form of the molecule, arising from the acidic mobile-phase solvents used in preparative HPLC. Helioamplexine N-Oxide (6). The corresponding helioamplexine N-oxide (6) was also proposed to be a major component in the nonreduced plant extract (Figure 2A) and a very minor component in the reduced plant extract (Figure 2B). This N-oxide was not isolated and only tentatively 8001

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry observed as a doublet in the 1H NMR at 1.31 ppm. These protons, H-3′ and CH3-4′, had a vicinal coupling of 6.4 Hz, an interaction supported by correlations present in the COSY spectrum. The COSY interactions between H-5′ and the adjacent methyl groups (C-6′ and C-7′) are consistent with the isopropyl of trachelanthic acid. Although H-5′ was observed as a broad multiplet at 2.05−2.07 ppm, it was coupled to H-6′ (δH 1.02, d) and H-7′ (δH 0.94, d), as observed in the COSY spectrum. Thus, it appeared that 7 was derived from indicine (3), the most abundant PA in H. amplexicaule. The NMR data obtained for compound 7 was also consistent with the presence of an angelyl ester33,34 rather than the alternate isomeric tiglic or senecioic acids (Figure 1).35 The distinct J couplings and chemical shifts of each of these acids allowed the unambiguous assignment of angelic acid with H-3″ δH 6.09 (qq, J = 7.4, 1.5 Hz) and the angelic methyl groups at δH 1.96 (dq, 7.4, 1.5 Hz, H-4″) and 1.82 (quintet, 1.5 Hz, H-5″)35,36 The assignment was confirmed by COSY interactions observed between H-3″ and both methyl groups H-4″ and H-5″. In the literature, 7-O-angelyl esters of retronecine are common,32,36−38 and if they are esterified at C-7, a downfield shift of H-7 would be expected. However, this was not observed in 7, with H-7 resonating at δ 4.47 as a broad singlet, comparable to the shifts observed in indicine (3) and helioamplexine (5). For comparison, the known 7-O-angelyl ester echiumine38 has a reported H-7 at δ 5.42. However, comparing the chemical shifts of H-3′ (Table 4), it is clear this is shifted downfield in 7 (δH 5.34) compared with in indicine (3, δH 4.03) and helioamplexine (5, δH 4.16). A literature value of δH 5.20 was reported for H-3′ in 3′-O-acetylindicine,29 and δH 5.20 was reported for H-3′ in scorpioidine, which contains a 3′-O-tiglyl moiety.39 Deshielding of the H-3′ proton is consistent with esterification at this position and with our observed MS/MS data.40 Crucially, H-3′ showed an HMBC correlation to C-1″ of the angelyl moiety, as observed in echivulgarine,37 suggesting that this angelyl group was attached at the C-3′ hydroxyl group. Thus, the structure of 7 was elucidated to be 3′-O-angelylindicine. Such a structure is analogous to the reassigned structure of anadoline and its Noxide, which have tiglyl esterification at C-3′.41 3′-O-Angelylindicine N-Oxide (8). The N-oxide 8 (corresponding to the free alkaloid 7) was also detected as a significant component in the nonreduced plant extract (Figure 2A) and also as a very minor component in the reduced plant extract (Figure 2B). This N-oxide was not isolated and was tentatively identified on the basis of HRAM LC-MS/MS showing C20H31NO7 + H+ (MH+) of m/z 398.2166 (see Table 2), with a mass fragmentation pattern characteristic of C-9 monoester N-oxides and a base peak at m/z 172.0967 (Table 2 and fragment ion A in Figure 3).27,28 By contrast, the more common 7-O-angelyl diester N-oxides displayed a dominant

peak at m/z 254.1387 (fragment ion B in Figure 3), as observed for the standards echimidine N-oxide and lasiocarpine N-oxide with our method. Hence, this N-oxide analogous to 7 is proposed to be esterified by angelic acid at C-3′. Esters at C-7 under mass-spectral conditions are much more stable than esters at C-9 because of the allylic nature of the C-9 hydroxyl group.32,42 5′-Hydroxyindicine (1). This alkaloid was present as a minor component in H. amplexicaule plant zinc-reduced extracts and observed by LCMS (Figure 2), but never isolated. It was first observed in associated honey extracts (vide infra), and retrospective HRAM LC-MS/MS analysis of the zincreduced plant extract demonstrated that this alkaloid was a very minor constituent in plant extracts (Figure 2). On the basis of mass-spectral fragmentation and biogenic considerations, this compound was proposed to be the 5′-hydroxy indicine derivative (1) with an observed MH+ 316.1750 in agreement with the formula C15H25NO6 + H+ (Table 2). Massspectral fragmentation revealed the predominant fragments at m/z 94.0655 (base peak), 120.0809, 138.0913, and 156.1018, characteristic of C-9 monoesters of retronecine27,28 (Table 2), and as such, the additional hydroxylation was deduced to be within the trachelanthic side chain of indicine. This alkaloid is therefore assumed to be a stereoisomer of leptanthine or echimiplatine,43,44 with an echimidinic acid type ester at C-9 but with the same stereochemistry as the parent alkaloid, indicine (3), the predominant alkaloid in H. amplexicaule. 3′-O-Angelylhelioamplexine (9). HRAM LC-MS/MS revealed a molecular ion MH+ at m/z 396.2372 corresponding to a molecular formula C21H33NO6 + H+. Mass-spectral fragmentation demonstrated the predominant fragments of m/ z 94.0656 (base peak), 120.0809, 138.0914, and 156.1019, which are again characteristic of C-9 monoesters of retronecine.27,28 A small peak at 296.1850 was observed, consistent with the intranecyl loss of angelic acid (100.0531 Da), with a further loss of 140.0835 yielding the core retronecine fragment of 156.1015 after the overall loss of a 3′-O-angelyl-homotrachelanthic acid fragment (C13H21O4). Hence it was inferred that this compound was 3′-O-angelyl helioamplexine (9). This minor compound (RT = 15.89 min) could not be characterized by NMR, because only trace amounts were collected, and thus the stereochemistry of the molecule is only inferred on biosynthetic grounds and on the basis of the structures of the co-occurring metabolites. N-Chloromethyl Artifacts of Isolation. In addition to expected pyrrolizidine alkaloids and associated N-oxides, a number of unexpected artifacts were isolated in this study, with three compounds (11−13) showing high-resolution isotope patterns consistent with the presence of a chlorine atom. NMR analysis revealed that these were N-chloromethyl artifacts of each of the major H. amplexicaule pyrrolizidine alkaloids, 3, 5, and 7 (Figure 4). LC-MS analysis demonstrated that these compounds were not present in the initial plant extracts (Figure 2) and that the chloromethylated compounds (11− 13) were only detected after the zinc-reduced PA extract was made alkaline with aqueous ammonia and extracted with dichloromethane prior to preparative HPLC separations. The formation of such chloromethyl artifacts with saturated pyrrolizidine alkaloids45−47 or with macrocyclic diester dehydropyrrolizidine alkaloids47 is not unprecedented, but to the best of our knowledge, simple mono- and diester dehydropyrrolizidine alkaloid artifacts such as 11−13 have not previously been identified. The extraction of tertiary

Figure 3. Fragment ions observed in (A) N-oxide monoesters (m/z 172.0967) and (B) N-oxide diesters with 7-O-angelyl esterification (m/z 254.1387). 8002

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

was observed at δH 5.36 (d, 10.3 Hz, H-10a) and 5.03 (H10b), with the chemical shift and splitting compatible with its attachment to the bridgehead nitrogen. Finally, in the 13C NMR spectra, the most significant differences observed between indicine (3) and the N-chloromethyl artifact (11) were those expected. The carbon atoms most affected (C-3, C5, C-8) were those adjacent to the newly functionalized bridgehead nitrogen (Tables 5 and 7), and an additional C-10 methylene at δC 68.5 appeared. N-Chloromethyl Helioamplexine (12). The second artifact was also subjected to HRAM LC-MS/MS analysis with a positive-ion M+ identified at m/z 362.1717/364.1688, corresponding to a molecular ion of C17H29ClNO5+ and exhibiting expected 35Cl−37Cl relative abundances. The fragmentation pattern (Table 3) displayed the loss of 158.0936 Da, equivalent to the α-cleavage of a homotrachelanthic acid fragment resembling that found in helioamplexine (5). In addition, 12 (C17H2935ClNO5+) also showed the base peak m/z 204.0781, with a loss of water giving 186.0676, which were again the N-chloromethyl analogues of the characteristic C-9 PA monoester fragments. Proton-NMR analysis of 12 (Table 6) started with the assignment of H-2 to the singlet at 5.85 ppm and H-8, observed as a broad singlet at 5.70 ppm. COSY interactions permitted the assignment of chloromethylene protons, δH 5.55 (d, J = 10.0 Hz, H-10a) and 5.07 (d, J = 10.0 Hz, H-10b) observed interacting solely with each other. Similarly, the H-9 methylene was assigned to the AB quartet at 5.16 and 4.97 ppm. H-7, at 5.07 ppm, was overlapping with the H-9b doublet. Centers adjacent to the nitrogen displayed significant downfield shifts. For instance, there was a 0.98 ppm difference between the H-8 in helioamplexine (5, 4.72 ppm) and the H-8 of artifact 12 (5.70 ppm). This difference is similar to that of the H-8 in indicine (3) and that of its N-chloromethyl artifact (11). The 13C NMR data for 12 are presented in Table 7. N-Chloromethyl 3′-O-Angelylindicine (13). An additional chloromethyl artifact, 13, was seen only in HRAM MS/ MS analysis (Table 3) and not further characterized. HRAM showed a molecular ion M+ at m/z 430.1978/432.1951 correlated to the molecular formula C21H33ClNO6+ (with predicted 35Cl−37Cl relative abundances), which is proposed to correspond to an N-chloromethyl artifact of 3′-Oangelylindicine (7). The fragmentation pattern of C21H3335ClNO6+, 13, displayed a loss of 100.0526 Da corresponding to the loss of angelic acid generating m/z 330.1452, with the base peak at m/z 204.0780 (derived from the 35Cl monoisotope) resulting from the sequential loss of angelyl and trachelanthic moieties attached to the necine base at C-9. Quantitation of Alkaloids in H. amplexicaule. HRAM LC-MS/MS quantitation of alkaloids present in the reduced extract of blue heliotrope plant against standard PAs, demonstrated the presence of 16.5 mg of total PAs per gram of dry weight, including 1.1 mg g−1 intermedine (2), 13.9 mg g−1 indicine (3), and 1.5 mg g−1 helioamplexine (5). (Helioamplexine was quantitated against the standard heliotrine, 10.) Blue Heliotrope and Commercial-Honey Profiles. Identification of the minor alkaloids present in H. amplexicaule provides a unique fingerprint of the plant’s HRAM LC-MS/ MS profile (Figure 2). When compared with pyrrolizidine alkaloid profiles from market honey samples, there is strong evidence that this plant species is being used as a floral source

Figure 4. N-chloromethyl artifacts 11−13 derived from major H. amplexicaule components.

aliphatic alkaloids with dichloromethane and trichloromethane has been reported to lead to the formation of similar unwanted quaternary salt artifacts with the general advice that such “solutions of the amines should not be stored for long periods”.48 In previous isolations of N-chloromethyl saturated and unsaturated pyrrolizidine alkaloids, these compounds have been reported as either natural plant components45 or artifacts.46,47,49 The basic conditions (pH ∼ 10) after the addition of aqueous ammonia would suggest that the PA would be nonprotonated, allowing it to act as a nucleophile in a concerted substitution reaction to generate these N-chloro methyl artifacts. The characterization of N-chloromethyl compounds 11−13 was accomplished via HRAMS LC-MS/ MS and a combination of 1D and 2D NMR (Tables 6 and 7). LC-MS/MS analysis also demonstrated that N-chloromethyl artifacts 11−13 each eluted from the preparative HPLC column after the corresponding parent alkaloid−N-oxide pair (3 and 4, 5 and 6, and 7 and 8). N-Chloromethyl Indicine (11). HRAM identified a positive-ion M+ of m/z 348.1562/350.1533 corresponding to a molecular formula of C16H27ClNO5+, with expected 35 Cl−37Cl relative abundance and a mass fragmentation pattern analogous to that seen in indicine (3). MS fragmentation of alkaloid 11 displayed a loss of 144.0781 Da, equivalent to a trachelanthic acid moiety, providing the base peak m/z 204.0781/206.0749 (the N-chloromethyl analogue of the indicine major fragment m/z 156.1015), and subsequent loss of H2O resulted in m/z 186.0676/188.0645 (the Nchloromethyl equivalent of indicine m/z 138.0911). This fragmentation pattern (Table 3) is consistent with the proposed N-chloromethyl indicine structure (11). The NMR data for the trachelanthic acid clearly indicated that the necic acid moiety remained unchanged during the generation of this artifact. Protons attached to tertiary carbons C-3′ and C-5′ appeared at δH 4.13 (q, H-3′) and 1.83 (septet H-5′), respectively. Methyl groups C-4′, C-6′, and C-7′ were observed at δH 1.21 (d, H-4′), 1.00 (d, H-6′), and 0.92 (d, H7′). In addition, the 1H NMR of the retronecine core of 11 was comparable to that of indicine N-oxide (4, Table 4), which was compatible with the presence of a quaternary nitrogen. H-8 (δ 5.57), which was adjacent to the nitrogen, experienced the most significant effects of this structural change, with shifts of 1.41 ppm relative to indicine (3, H-8, 4.16) and 0.47 ppm relative to indicine N-oxide (4, δ 5.10). The H-7 proton was observed at δH 4.96 (d, H-7), which was again comparable to indicine N-oxide (δH 4.77, H-7). The diastereomeric protons at C-9 were observed at δH 5.08 (d, H-9a) and 5.02 (d, H-9b). Assignment of the methylene protons of the chloromethyl group was assisted by COSY, through which this AB quartet 8003

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

Figure 5. HRAM LC-MS/MS traces of pyrrolizidine alkaloids for (A) H. amplexicaule compared with (B−D) traces for three commercial honey samples from different sources containing 1.4−2.0 μg of PAs per kilogram of honey and (E) that of isolated helioamplexine.

States), India, Italy, South Africa, Algeria, Egypt, and Australia.50 Within Australia, it is particularly abundant in the eastern and southeastern parts of Australia, including Queensland, where the samples in this study were sourced. Indicine (3) is not always included in targeted LC-MS/MS surveys of honey,5−7 and the honey alkaloid profiles shown in Figure 5 highlight the need to include this alkaloid in such surveys, particularly in regions where H. amplexicaule is abundant. Honey as a Dietary Source of Pyrrolizidine Alkaloids. The honey samples shown in Figure 5 were selected to demonstrate that the indicine-dominant H. amplexicaule profile did occur in honey from this region, and the depicted results do not represent an extensive survey of such honeys. However, consideration of the measured pyrrolizidine alkaloid levels in these individual samples suggests that a more extensive survey of Australian honey is warranted. The cumulative toxicity of the 1,2-unsaturated PAs has been demonstrated in experimental animals and is characterized by hepatotoxicity, developmental toxicity, genotoxicity, and carcinogenicity.1 As a result, provisional tolerable daily intakes (PTDIs) have been recommended to limit the human consumption of such alkaloids, with honey being only one of several considered dietary sources.3,4,51,52 These PTDI values (which vary from 1 to 0.007 μg of PA per kilogram of body weight per day), when combined with estimates of food consumption from nutritional-survey data, enable assessment of dietary PA exposure. According to Australian nutritional surveys, children have a higher honey intake per kilogram of body weight, with the top 5% consuming at least 28.6 g of honey per day for infants 2−4 years with an average body weight of 17 kg.53,54 Consumption of H-PA#14, which contains 2.0 μg of PAs per gram of honey, by this vulnerable subgroup would provide a daily PA intake of 3.3 μg of PA per kilogram of body weight per day, exceeding all recommended PTDI values. In past decades, honey foraged from PA-containing plants has been reported to contain up to 3.9 μg of PAs per gram of honey.15,55,56 There have, however, been significant efforts to restrict the PA content of honey in

for honey by bees (Figure 5). Honey samples such as H-PA#141, H-PA#101, and H-PA#14 (Figure 5) were purchased from commercial suppliers, and clearly demonstrated not only the presence of the H. amplexicaule major alkaloid indicine (3) but also the lesser but perhaps more uniquely characteristic minor components 5′-hydroxyindicine (1), intermedine (2), helioamplexine (5), and 3′-O-angelylindicine (7). It is this complete profile or fingerprint that enables PA-floral-source identification for these honey samples. Importantly, heliotrine (10) was not detected in either the H. amplexicaule plant extracts or in these honey extracts. Of the 30 standards of pyrrolizidine alkaloids analyzed in our standard honey screen (Table 1), only indicine (3) and intermedine (2) and traces of the corresponding N-oxides were detected in these honey samples. Each of the honeys presented in Figure 5 was obtained commercially, and none attributed the specific floral source. However, when compared with the plant alkaloid profile (Figure 5), HRAM LC-MS/MS analysis indicated that H. amplexicaule was a significant contributing floral source for these honeys. One was called simply organic honey (H-PA#141), one was attributed to “ground flora” (HPA#101), and the other was attributed to a specific suburban Brisbane location (H-PA#14). These honeys were calculated to contain between 1.4 and 2.0 μg of PAs per gram of honey by HRAM LC-MS/MS, with indicine (3) representing approximately 80% of the measured alkaloid content (Figure 5). It is worth noting that the detected pyrrolizidine alkaloids in the H. amplexicaule plant are present primarily as the N-oxides (>90%), whereas the honeys contain primarily the free alkaloids (>99%). It is assumed that these plant-derived Noxides are being reduced within the bee digestive system, as has been recorded for other such alkaloids.12 H. amplexicaule is clearly attractive to bees as a nectar source, and the potential contamination of honey should be considered in all areas where this plant is present. This species is native to regions of central South America, but it has become naturalized in diverse regions, including Alabama (United 8004

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry

(2) Fletcher, M. T.; McKenzie, R. A.; Blaney, B. J.; Reichmann, K. G. Pyrrolizidine alkaloids in Crotalaria taxa from northern Australia: risk to grazing livestock. J. Agric. Food Chem. 2009, 57, 311−319. (3) EFSA Panel on Contaminants in the Food Chain (CONTAM). Scientific opinion on pyrrolizidine alkaloids in food and feed. EFSA J. 2011, 9, 2406. (4) EFSA Panel on Contaminants in the Food Chain (CONTAM). Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions and food supplements. EFSA J. 2017, 15, 4908. (5) Griffin, C. T.; O’Mahony, J.; Danaher, M.; Furey, A. Liquid chromatography tandem mass spectrometry detection of targeted pyrrolizidine alkaloids in honeys purchased within Ireland. Food Analytical Methods 2015, 8, 18−31. (6) Dubecke, A.; Beckh, G.; Lullmann, C. Pyrrolizidine alkaloids in honey and bee pollen. Food Addit. Contam., Part A 2011, 28, 348− 358. (7) Griffin, C. T.; Mitrovic, S. M.; Danaher, M.; Furey, A. Development of a fast isocratic LC-MS/MS method for the highthroughput analysis of pyrrolizidine alkaloids in Australian honey. Food Addit. Contam., Part A 2015, 32, 214−228. (8) Edgar, J. A.; Roeder, E.; Molyneux, R. J. Honey from plants containing pyrrolizidine alkaloids: A potential threat to health. J. Agric. Food Chem. 2002, 50, 2719−2730. (9) Kempf, M.; Reinhard, A.; Beuerle, T. Pyrrolizidine alkaloids (PAs) in honey and pollen-legal regulation of PA levels in food and animal feed required. Mol. Nutr. Food Res. 2010, 54, 158−68. (10) Lucchetti, M. A.; Glauser, G.; Kilchenmann, V.; Dübecke, A.; Beckh, G.; Praz, C.; Kast, C. Pyrrolizidine alkaloids from Echium vulgare in honey originate primarily from floral nectar. J. Agric. Food Chem. 2016, 64, 5267−5273. (11) Yang, M.; Ruan, J.; Gao, H.; Li, N.; Ma, J.; Xue, J.; Ye, Y.; Fu, P. P.-C.; Wang, J.; Lin, G. First evidence of pyrrolizidine alkaloid Noxide-induced hepatic sinusoidal obstruction syndrome in humans. Arch. Toxicol. 2017, 91, 3913−3925. (12) Reinhard, A.; Janke, M.; von der Ohe, W.; Kempf, M.; Theuring, C.; Hartmann, T.; Schreier, P.; Beuerle, T. Feeding deterrence and detrimental effects of pyrrolizidine alkaloids fed to honey bees (Apis mellifera). J. Chem. Ecol. 2009, 35, 1086. (13) Gottschalk, C.; Huckauf, A.; Dübecke, A.; Kaltner, F.; Zimmermann, M.; Rahaus, I.; Beuerle, T. Uncertainties in the determination of pyrrolizidine alkaloid levels in naturally contaminated honeys and comparison of results obtained by different analytical approaches. Food Addit. Contam., Part A 2018, 35, 1366− 1383. (14) Culvenor, C. C. J.; Edgar, J. A.; Smith, L. W. Pyrrolizidine alkaloids in honey from Echium plantagineum L. J. Agric. Food Chem. 1981, 29, 958−60. (15) Beales, K. A.; Betteridge, K.; Colegate, S. M.; Edgar, J. A. Solidphase extraction and LC-MS analysis of pyrrolizidine alkaloids in honeys. J. Agric. Food Chem. 2004, 52, 6664−6672. (16) Julien, M. H. Biological control of rangeland weeds in Australia. Rangeland Journal 2006, 28, 47−54. (17) Blake, S. T.; Roff, C. The honey flora of Queensland, 3rd ed.; Queensland Department of Primary Industries Information Series Q187015; Queensland Deptartment of Primary Industries: Brisbane, 1988. (18) Blake, S. T.; Roff, C. Honey grades, 2013. Queensland Government.https://www.daf.qld.gov.au/business-priorities/ biosecurity/animal-biosecurity-welfare/animal-health-pests-diseases/ beekeeping-in-queensland/hive-management/honey-grades (accessed Jan 16, 2019). (19) Heliotropium amplexicaule Vahl. In Weeds of Australia, Biosecurity Queensland Edition; Queensland Government, 2016. https://keyserver.lucidcentral.org/weeds/data/media/Html/ heliotropium_amplexicaule.htm (accessed Jan 16, 2019). (20) Briese, D. Heliotropium amplexicaule Vahlblue heliotrope. In Biological Control of Weeds in Australia; Julien, M. H., McFadyen, R. E.

recent years through improved positioning of honey hives to avoid culprit floral sources. Recent social trends of keeping bee hives within urban settings is, however, problematic in this regard, because of the reduced control of PA-containing plants in urban environments. In our study, honey sourced from a suburban location (H-PA#14) was measured to contain 2.0 μg of PAs per gram of honey. This honey was sold through Health Food Shops and is unlikely to be blended (or analyzed for PA content), increasing the exposure risk for unsuspecting consumers seeking a more “natural” alternative. These results confirm the need for all honey producers to be aware of H. amplexicaule as a potential PA source, most particularly in small- to medium-scale productions where honey is sourced from a single location.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jafc.9b02136.



HPLC-ELSD chromatogram of PA-enriched crude extract and original 1D and 2D NMR spectra of PAenriched crude extract (predominantly indicine, 3), protonated indicine (3H+), indicine N-oxide (4), helioamplexine (5), N-chloromethyl indicine (11), Nchloromethyl helioamplexine (12), and a mixture of intermedine (2) and 3′-O-angelylindicine (7) (PDF)

AUTHOR INFORMATION

Corresponding Author

*Tel.: +61 7 3443 2479. E-mail: mary.fl[email protected]. ORCID

Matheus Carpinelli de Jesus: 0000-0003-0889-8065 Joanne T. Blanchfield: 0000-0003-1338-7446 Mary T. Fletcher: 0000-0003-0189-3376 Author Contributions ⊥

M.C.J. and N.L.H. contributed equally to this work.

Funding

This work was supported by Queensland Health Grant RSS17002. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We would like to thank Greg Pierens at the Center of Advanced Imaging (CAI) for providing excellent technical assistance with the 700 MHz NMR experiments.



ABBREVIATIONS USED DCM, dichloromethane; ELSD, evaporative light-scattering detector; HPLC, high-performance liquid chromatography; HRAM, high-resolution accurate mass; LC-MS, liquid chromatography−mass spectrometry; LC-MS/MS, liquid chromatography−tandem mass spectrometry; PA, pyrrolizidine alkaloid; PTDI, provisional tolerable daily intake



REFERENCES

(1) Fu, P. P.; Xia, Q.; Lin, G.; Chou, M. W. Pyrrolizidine alkaloids Genotoxicity, metabolism enzymes, metabolic activation, and mechanisms. Drug Metab. Rev. 2004, 36, 1−55. 8005

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006

Article

Journal of Agricultural and Food Chemistry C., Cullen, J. M., Eds.; CSIRO Publishing: Clayton, VIC, Australia, 2012; pp 282−288. (21) Ketterer, P. J.; Glover, P. E.; Smith, L. W. Blue heliotrope (Heliotropium amplexicaule) poisoning in cattle. Aust. Vet. J. 1987, 64, 115−117. (22) Dellow, J. J.; Bourke, C. A.; McCaffery, A. C. Blue heliotrope; Primefact 653; New South Wales Department of Primary Industries, 2008. http://www.dpi.nsw.gov.au/__data/assets/pdf_file/0003/ 233355/Blue-heliotrope.pdf (accessed Jan 16, 2019). (23) Hartmann, T.; Theuring, C.; Beuerle, T.; Klewer, N.; Schulz, S.; Singer, M. S.; Bernays, E. A. Specific recognition, detoxification and metabolism of pyrrolizidine alkaloids by the polyphagous arctiid Estigmene acrea. Insect Biochem. Mol. Biol. 2005, 35, 391−411. (24) Jones, A.; Culvenor, C.; Smith, L. Pyrrolizidine alkaloids: A carbon-13 N.M.R. study. Aust. J. Chem. 1982, 35, 1173−1184. (25) Ogawa, T.; Niwa, H.; Yamada, K. An efficient enantioselective synthesis of indicine N-oxide, an antitumor pyrrolizidine alkaloid. Tetrahedron 1993, 49, 1571−1578. (26) Zalkow, L. H.; Glinski, J. A.; Gelbaum, L. T.; Fleischmann, T. J.; McGowan, L. S.; Gordon, M. M. Synthesis of pyrrolizidine alkaloids indicine, intermedine, lycopsamine, and analogues and their N-oxides. Potential antitumor agents. J. Med. Chem. 1985, 28, 687− 694. (27) These, A.; Bodi, D.; Ronczka, S.; Lahrssen-Wiederholt, M.; Preiss-Weigert, A. Structural screening by multiple reaction monitoring as a new approach for tandem mass spectrometry: presented for the determination of pyrrolizidine alkaloids in plants. Anal. Bioanal. Chem. 2013, 405, 9375−9383. (28) Avula, B.; Sagi, S.; Wang, Y.-H.; Zweigenbaum, J.; Wang, M.; Khan, I. A. Characterization and screening of pyrrolizidine alkaloids and N-oxides from botanicals and dietary supplements using UHPLChigh resolution mass spectrometry. Food Chem. 2015, 178, 136−148. (29) Ogihara, K.; Miyagi, Y.; Higa, M.; Yogi, S. Pyrrolizidine alkaloids from Messerschmidia argentea. Phytochemistry 1997, 44, 545− 547. (30) Roeder, E.; Breitmaier, E.; Birecka, H.; Frohlicht, M. W.; Badzies-Crombach, A. Pyrrolizidine alkaloids of Heliotropium spathulatum. Phytochemistry 1991, 30, 1703−1706. (31) Rycroft, D. S.; Stirling, I. R.; Robins, D. J. Assignment of the 1H and 13C NMR spectra and conformational analysis of the pyrrolizidine alkaloid 13-O-acetyldicrotaline. Magn. Reson. Chem. 1992, 30, S42−S45. (32) Colegate, S. M.; Gardner, D. R.; Joy, R. J.; Betz, J. M.; Panter, K. E. Dehydropyrrolizidine alkaloids, including monoesters with an unusual esterifying acid, from cultivated Crotalaria juncea (Sunn Hemp cv.’Tropic Sun’). J. Agric. Food Chem. 2012, 60, 3541−50. (33) Fraser, R. R. Long-range coupling constants in the N.M.R. spectra of olefines. Can. J. Chem. 1960, 38, 549−553. (34) Joseph-Nathan, P.; Wesener, J. R.; Günther, H. A twodimensional NMR study of angelic and tiglic acid. Org. Magn. Reson. 1984, 22, 190−191. (35) Roeder, E.; Wiedenfeld, H.; Schraut, R. Pyrrolizidine alkaloids from Alkanna tinctoria. Phytochemistry 1984, 23, 2125−2126. (36) Roitman, J. N. Longitubine and neolatifoline, new pyrrolizidine alkaloids from Hackelia longituba. Aust. J. Chem. 1988, 41, 1827. (37) Cairns, E.; Hashmi, M. A.; Singh, A. J.; Eakins, G.; Lein, M.; Keyzers, R. Structure of echivulgarine, a pyrrolizidine alkaloid isolated from the pollen of Echium vulgare. J. Agric. Food Chem. 2015, 63, 7421−7427. (38) Stermitz, F. R.; Pass, M. A.; Kelley, R. B.; Liddell, J. R. Pyrrolizidine alkaloids from Cryptantha species. Phytochemistry 1993, 33, 383−387. (39) Resch, J. F.; Rosberger, D. F.; Meinwald, J.; Appling, J. W. Biologically active pyrrolizidine alkaloids from the True Forget-MeNot, Myosotis scorpioides. J. Nat. Prod. 1982, 45, 358−362. (40) Fleming, I.; Williams, D. H. Spectroscopic methods in organic chemistry, 5th ed.; McGraw-Hill: Berkshire, England, 1995.

(41) Culvenor, C. C. J.; Edgar, J. A.; Frahn, J. L.; Smith, L. W.; Ulubelen, A.; Doganca, S. The structure of anadoline. Aust. J. Chem. 1975, 28, 173. (42) Pedersen, E.; Larsen, E. Mass spectrometry of some pyrrolizidine alkaloids. Org. Mass Spectrom. 1970, 4, 249−256. (43) Mroczek, T.; Ndjoko, K.; Glowniak, K.; Hostettmann, K. Online structure characterization of pyrrolizidine alkaloids in Onosma stellulatum and Emilia coccinea by liquid chromatography-ion-trap mass spectrometry. J. Chromatogr. A 2004, 1056, 91−97. (44) Skoneczny, D.; Weston, P. A.; Zhu, X.; Gurr, G. M.; Callaway, R. M.; Weston, L. A. Metabolic profiling of pyrrolizidine alkaloids in foliage of two Echium spp. invaders in Australia – A case of novel weapons? Int. J. Mol. Sci. 2015, 16, 26721−37. (45) Shi, B.-J.; Xiong, A.-Z.; Zheng, S.-S.; Chou, G.-X.; Wang, Z.-T. Two new pyrrolizidine alkaloids from Senecio nemorensis. Nat. Prod. Res. 2010, 24, 1897−1901. (46) Chen, L.; Huang, S.; Li, C. Y.; Gao, F.; Zhou, X. L. Pyrrolizidine alkaloids from Liparis nervosa with antitumor activity by modulation of autophagy and apoptosis. Phytochemistry 2018, 153, 147−155. (47) Pérez-Castorena, A.-L.; Arciniegas, A.; Pérez, R.; Gutierrez, H.; Toscano, R. A.; Villaseñor, J. L.; Romo de Vivar, A. Iodanthine, a pyrrolizidine alkaloid from Senecio iodanthus and Senecio bracteatus. J. Nat. Prod. 1999, 62, 1039−1043. (48) Almarzoqi, B.; George, A. V.; Isaacs, N. S. The quarternisation of tertiary amines with dihalomethane. Tetrahedron 1986, 42, 601− 607. (49) Pérez-Castorena, A.-L.; Arciniegas, A.; Alonso, R. P.; Villaseñor, J. L.; Romo de Vivar, A. Callosine, a 3-alkyl-substituted pyrrolizidine alkaloid from Senecio callosus. J. Nat. Prod. 1998, 61, 1288−1291. (50) Heliotropium amplexicaule Vahl. Plants of the World online, Kew Science. http://powo.science.kew.org/taxon/urn:lsid:ipni. org:names:116657-1 (accessed Jan 16, 2019). (51) European Food Safety Authority (EFSA). Dietary exposure assessment to pyrrolizidine alkaloids in the European population. EFSA J. 2016, 14, e04572. (52) Pyrrolizidine alkaloids in food: A toxicological review and risk assesment; Technical Report Series No. 2; Australian New Zealand Food Authority, 2001. (53) Information Sheet - Pyrrolizidine Alkaloids (PAs). Australian Honey Bee Industry Council (AHBIC). https://www.honeybee.org.au/ pdf/INFORMATION_SHEET_on_PAs.pdf (accessed Jan 16, 2019). (54) National Nutrition Survey: Foods Eaten, Australia, 1995; ABS Catalogue No. 4804.0; Australian Bureau of Statistics, 1995. https:// www.abs.gov.au/AUSSTATS/[email protected]/productsbyCatalogue/ 9A125034802F94CECA2568A9001393CE?OpenDocument. (55) Deinzer, M. L.; Thomson, P. A.; Burgett, D. M.; Isaacson, D. L. Pyrrolizidine alkaloids: their occurrence in honey from tansy ragwort (Senecio jacobaea L.). Science 1977, 195, 497−9. (56) Betteridge, K.; Cao, Y.; Colegate, S. M. Improved method for extraction and LC-MS analysis of pyrrolizidine alkaloids and their Noxides in honey: application to Echium vulgare honeys. J. Agric. Food Chem. 2005, 53, 1894−902.

8006

DOI: 10.1021/acs.jafc.9b02136 J. Agric. Food Chem. 2019, 67, 7995−8006