A fragmentation study on four C19-diterpenoid alkaloids by electrospray ionization ion-trap time-of-flight tandem mass spectrometry

High-resolution electrospray ionization ion-trap time-of-flight tandem mass spectrometry (HR-ESI-IT-TOF-MSn) in positive-ion mode was used to determine the accurate masses and fragmentation pathways of four C19-diterpenoid alkaloids, aconitine (1), yunnaconitine (2), crassicauline A (3), and benzoylmesaconine (4). The [M+H]+ ions of compounds 1–4 were readily observed in conventional single-stage mass spectrometry. Based on the MS1–6 analyses, detailed fragmentation rules of the four compounds were proposed. The neutral losses of AcOH, MeOH, H2O, CO, C2H4, PhCOOH and p-OMePhCOOH segments were the characteristic eliminations from the precursor ions due to the presence of acetyl, methoxyl, hydroxyl, N-ethyl, benzoyl and p-methoxyl-benzoyl units in the structures. Benefited from the high resolution of the mass analyzer, the loss of 28 Da corresponding to CO or CH4 segment in product ions was unambiguously distinguished. The losing sequence of the main substituent groups was summarized as: C(8)-acetyl>C(16)-methotyl>C(15)-hydroxyl>C(6)-methoxyl>C(1)-methoxyl/C(3)-hydroxyl>C(18)-methoxyl>>C(13)-hydroxyl. The sequential loss of (16)-methoxyl moiety and CO (generating from enol–ketone tautomerism) groups could be recognized as the characteristic eliminations for the compounds with C(16)-methoxyl and C(15)-hydroxyl groups simultaneously. The application of HR-ESI-IT-TOF-MSn technique to investigate the fragmentation of C19-diterpenoid alkaloids provided useful information to understand their fragmentation behaviors.


Introduction
C 19 -diterpenoid alkaloids as the main substance of plants Aconitum L. (Ranunculaceae) show wide range of bioactivities [1 -3], such as anti-inflammatory, cardiotonic, and analgesic effects; however, the obvious toxicity limits their applications in clinic [4]. According to the previous reports [4], diester-diterpenoid alkaloids are the major constituents of the root of Aconitum kusnezoffii Reichb. and cause toxicities and side effects. Therefore, many strategies were attempted to reduce the toxicity or establish a fast and effective method for detecting diester-diterpenoid alkaloids [5 -8].
Mass spectrometry (MS) with high sensitivity and resolution has become the routine method in various aspect of analytic chemistry. In particular, the tandem mass spectrometry (MS n ) techniques make it possible to determine the relationship between the precursor and product ions, by which the fragmentation rules can be resolved. Cui et al. [9] investigated the mass spectrometric behavior of eight phenanthroindolizidine alkaloids by ESI-MS n and established a fast method by the rules to identify the tracelevel alkaloids in the crude of Tylophora atrofolliculata. Nagy et al. [10] reported the detailed fragmentation behavior of protonated noscapines under electrospray condition. These studies have demonstrated the advantages of using the MS/ MS technique for identifying natural products. Although some papers reported the fragmentation rules of C 19 -diterpenoid alkaloids [11 -15], the MS n analyses were performed by low-resolution mass analyzer and the investigation based on high resolution mass analyzer were scarce.
Aconitine (1), yunnaconitine (2), crassicauline A (3), and benzoylmesaconine (4) (as shown in Figure 1) are four aconitinetype alkaloids with a C 19 skeleton, of which aconitine (1), yunnaconitine (2), and crassicauline A (3) belong to the diesterditerpenoid alkaloids. Compounds 1-4 are appropriate candidates for tandem mass spectrometry (MS n ) study due to the suitable molecule masses (.500 Da) and the prolific substituent groups. To our knowledge, there are no systematic reports about the high-resolution MS n fragmentation rules of the four compounds. In this paper, the LC/MS-IT-TOF mass spectrometer equipped with an electrospray ionization source linked to ion-trap and time-of-flight mass analyzers (ESI-IT-TOF) which enables fast acquisition of multistage tandem spectrometry (MS 1210 ) with high resolution [16,17] was used to investigate the fragmentation rules of compounds 1-4. It is important to use HR-ESI-IT-TOF-MS n technique to investigate the fragmentation of C 19 -diterpenoid alkaloids, which will provide useful information to understand their fragmentation behaviors.

Results and discussion
Aconitine (1), yunnaconitine (2), crassicauline A (3), and benzoylmesaconine (4) are four natural C 19 -diterpenoid alkaloids with alkalinity so that MS n experiments in positive mode were applied to characterize their fragmentation behaviors. The proposed fragmentation pathways are shown in Figures 2, 3, S2 and S3 and data for accurate masses and elemental compositions from tandem mass spectrometry are shown in Tables 1-4.

ESI-MS n results
2.1.1. ESI-IT-TOF MS n fragmentations of aconitine (1) in positive mode (as shown in Figure 2 and Table 1) The full-scan mass spectrometry for aconitine (1) in positive mode was analyzed to afford the [M þ H] þ ion (1a) at m/z 646.3220, corresponding to the molecular formula C 34 H 48 NO 11 . Ion 1a was selected as precursor ion to yield versatile product ions (1b-1k) in MS 2 experiment, and the [M þ H 2 AcOH] þ ion at m/z 586 (1b) with high intensity (90%) further yielded the fragment at m/z 554 (1c) due to the departure of MeOH at C(16) position. The ketone-enol tautomers [between D 15,16 and C(15)-hydroxyl group] of the unstable ion 1c lost a carbonyl moiety as one molecule of CO to give rise to the ion 1d (90%) at m/z 526 or eliminated C(6)-methoxyl unit directly to produce the ion 1e at m/z 522. Then the ion 1d lost C(6)-methoxyl group to yield the ion at m/z 494 (1f, 100%). After that, 1e and 1f both lost the methoxyl moiety at C(1) position to generate ions 1g (m/z 490) and 1i (m/z 462), respectively. The ion 1h at m/z 476 could be interpreted by the departure of C(3)-hydroxyl group from 1f, and the elimination of a 122 Da (PhCOOH) from 1d and 1g was happened to afford ions 1j (m/z 404) and 1k (m/z 368) due to the presence of C(14)-benzoyl unit.
Ions 1b (m/z 586), 1h (m/z 476), 1j (m/z 404), and 1k (m/z 368) were selected as precursor ions to perform MS 3 experiments, which provided ions 1c, 1d, 1f, 1h, 1i, 1j, and 1l -1q. The fragments 1c, 1d, 1f, 1h, 1i, and 1j could be also detected in the MS 2 analysis. The ion 1l (m/z 354, 100%) was the most abundant product ion due to the loss of PhCOOH from precursor ion 1h. Precursor ion 1j eliminated C(6)-methoxl group to generate the ion 1m at m/z 372 due to the formation of a stably conjugated structure. Then 1m gave rise to ions 1l (m/z 354) and 1n (m/z 340) in agreement with the departure of C(3)-hydroxyl and C(1)-methoxyl units, and the ion 1o (m/z  322) could be interpreted by eliminating C (1)-methoxyl group from 1l or C(3)hydroxyl moiety from 1n. Besides 1l, 1n, and 1o, the ion 1p (m/z 218) was detected with high abundance, which could be interpreted by the loss of C 7 H 4 O segment from 1o. In another route, the precursor ion 1k produced the ion 1n (m/z 340) by losing one molecule of CO and further gave rise to the ion 1q (m/z 296), which was attributed to the departure of C 2 H 4 O segment at C(4) position.
In MS 4 scan of 1f at m/z 494, two product ions, 1h at m/z 476 and 1i at m/z 462 (100%), were observed. At the same time ion, 1l at m/z 354 was selected as precursor ion to generate a fragment at m/z 322 (1o), which was assigned to the elimination of C(1)-methoxyl moiety on 1l.

ESI-IT-TOF MS n fragmentations
of yunnaconitine (2) in positive mode (as shown in Figure S2 and Table 2) Unlike aconitine (1), yunnaconitine (2) showed simple fragmentation characteristics relatively. In the single-stage mass spectrometry of yunnaconitine (2) in positive mode, the [M þ H] þ ion at m/z 660.3405 (2a) was found and its molecular formula was C 35 H 50 NO 11 . In MS 2 analysis, 2a produced ions 2b-2i in which 2b (m/z 600, 100%) and 2c (m/z 568, 70%) with high abundance were assigned to the elimination of C(8)-acetyl and C(16)methoxyl units just as the fragmentation rules of 1b and 1c in aconitine (1). Then 2c lost C(3)-hydroxyl or C(6)-methoxyl group to generate ions 2d (m/z 550, 100%) and 2e (m/z 536). The ion 2f (m/z 518) arose from two originations, 2d losing C(6)-methoxyl group and 2e losing C(3)-hydroxyl unit. Fragments 2g at m/z 486 and 2h at m/z 366 both derived from the ion 2f by losing C(6)methoxyl unit for the former and eliminating C(14)-p-methoxyl-benzoyl group for the latter. Then 2g lost C(14)-p-methoxylbenzoyl unit and 2h split C(1)-methoxyl group to generate the same product ion 2i at m/z 334. Precursor ion 2b (m/z 600) was analyzed in MS 3 experiment to afford product ions 2c -2i in which ion 2j at m/z 447 had a none-nitrogen molecular formula as C 27 H 27 O 6 that predicted the nitrogen heterocyclic ring on 2d split to lose a C 5 H 13 NO segment. The ion 2i was also observed when 2h (m/z 366) acted as precursor ion due to the departure of C(1)methoxyl group.
In MS 4 analysis, the product ions 2f -2j were detected from precursor ion 2d (m/ z 550) just as above-mentioned interpretations. Another precursor ion 2f (m/z 518) afforded four fragments, 2g at m/z 486 (100%), 2k at m/z 472, 2h at m/z 366, and 2i at m/z 334, in which 2k was arose from the elimination of C(4)-methoxymethyl unit as a C 2 H 6 O segment from 2f.
Via MS 5 analysis, the ion 2l at m/z 454 was observed from precursor ion 2g (m/z 486) attributed to the loss of C(18)methoxyl group. Precursor ion 2i (m/z 334) gave rise to three fragments, the ion 2m at m/z 302, 2n at m/z 290, and 2o at m/z 288, based on losing C(18)-methoxyl group or directly eliminating C(4)-methoxymethyl unit as C 2 H 4 O or C 2 H 6 O segment, respectively.
In the MS n analyses of yunnaconitine (2), the losses of C(8), C(16), C(3), C(1)moieties gave rise to a series of ions with high intensity such as 2b, 2c, 2d, and 2g which were similar as aconitine (1). Figure S3 and Table 3) The first-stage mass spectrometry of crassicauline A (3) generated the [M þ H] þ ion at m/z 644.3476 (3a), corresponding to the molecular formula C 35 H 50 NO 10 . For MS 2 analysis of 3a (m/z 644), ions 3b at m/z 584 (76%), and 3c at m/z 552 (100%) with high intensity were observed and assigned with the losses of C (8)-acetyl and C(16)-methoxyl groups sequentially from 3a. Because of the absence of C(3)-hydroxyl group, the ion 3c lost N-ethyl unit easily to produce the ion 3d (m/z 524) and then 3d lost C(6)methoxyl group to yield the ion 3f at m/z 492. Besides, the ion 3c afforded the ion 3e at m/z 520 with high abundance (76%) relatively by eliminating C(6)-methoxyl unit to form a stable conjugated structure. The ion 3h at m/z 400 was detected readily corresponding to the elimination of C(14)p-methoxyl-benzoyl group from 3c, too, and 3h further lost C(6)-methoxyl unit to afford the ion 3i at m/z 368. Certainly, the loss of N-ethyl or C(14)-p-methoxylbenzoyl unit from 3e was the pathway to generate the fragments 3f and 3i as well. 3e also gave rise to the ion 3g at m/z 488 by eliminating C(1)-methoxyl group. Thus the ion 3j (m/z 336) could be explained by the loss of C(14)-p-methoxyl-benzoyl unit from 3g or the loss of C(1)-methoxyl group from 3i, and then 3j eliminated one molecule of MeOH at C(18) position to produce the ion 3k at m/z 304.

ESI-IT-TOF MS n fragmentations of crassicauline A (3) in positive mode (as shown in
In agreement with MS 2 analysis, the ions 3c -3k and 3f, 3g, 3i, 3j were all found when 3b (m/z 584), 3e (m/z 520), and 3h (m/z 400) were selected as precursor ions in MS 3 experiments. Ions 3i, 3j, and 3l-3q were obtained in the MS 3 analysis of 3h in which 3l -3q were interpreted as follows: the ion 3h (m/z 400) eliminated N-Et or C(4)-methoxymethyl group to generate the ions 3l (m/z 372) and 3m (m/z 354), and then 3l and 3m deducted C(6)-methoxyl unit relatively to yield the ions 3n (m/z 340) and 3p (m/z 322); the ion 3o (m/z 324) derived from 3i due to the loss of C(4)-methoxymethyl group as C 2 H 4 O segment; the base peak ion 3q (m/z 308) was formed by losing N-ethyl unit from 3j directly. Whilst, 3r (m/z 276) was the base peak ion in the MS 3 analysis of precursor ion 3k which was owed to the elimination of N-ethyl group as C 2 H 4 unit.
The MS 4 analysis of four precursor ions, 3c (m/z 552), 3g (m/z 488), 3i (m/z 368), and 3j (m/z 336), yield nine product ions, 3d -3j, 3k, 3q, and 3r -3t, in which the fragmentation pathways of 3d -3j, 3q, and 3k had been elucidated in MS 2 and MS 3 experiments. The ion 3r at m/z 276 could be explained by the loss of MeOH from 3q at C(18) position or the elimination of C 2 H 4 segment from 3k on nitrogen-atoms. The ion 3s was attributed to the loss of tertiary carbon as CH 2 group at C(4) position from 3k. Then 3s gave rise to the ion 3t by eliminating N-ethyl group as C 2 H 4 unit. Another origination of 3t was the ion 3r, which lost the tertiary carbon moiety as CH 2 segment at C(4) position to generate fragment 3t.
In MS 5 analysis of precursor ion 3d (m/ z 524), the ion 3f (m/z 492) was detected due to the elimination of C(6)-methoxyl unit and then 3f generated the ion 3u (m/z 460) owing to the loss of C(1)-methoxyl group.
To study the fragmentation route of the ion 3f (m/z 492) in depth, the MS 6 analysis of ion 3f was carried out from which product ion 3u at m/z 460 was detected and attributed to the loss of C(1)-methoxyl unit.

ESI-IT-TOF MS n fragmentations
of benzoylmesaconine (4) in positive mode (as shown in Figure 3 and Table 4) It was easy to detect the [M þ H] þ ion (4a) at m/z 590.2966 of benzoylmesaconine (4) in the positive full-scan mass spectrometry concurring with the molecular formula C 31 H 44 NO 10 . A series of ions (4b-4j) were found when 4a acted as precursor ion in MS 2 experiment. The loss of C(8)hydroxyl or C(16)-methoxyl unit from 4a leaded to the appearance of the ions 4b at m/z 572 and 4c at m/z 558 relatively. The ion 4d (m/z 540, 100%) could be explained by the elimination of C(16)-methoxyl group on 4b or C(8)-hydroxyl unit on 4c and then 4d generated the ion 4f at m/z 522 by losing C(3)-hydroxyl. Ions 4c, 4d, and 4f all eliminated one molecule of MeOH to produce the [P-MeOH] þ (P: product ion) ions, 4e at m/z 526, 4g at m/z 508 (90%), and 4i at m/z 490 due to the existence of C (6)-methoxyl group, respectively. The ions 4e and 4g gave rise to the ions 4h (m/z 494) and 4j (m/z 476) relatively due to the further splitting of C(1)-methoxyl unit. Furthermore, the ion 4e generated the ion 4g by eliminating C(8)-hydroxyl group, as well as 4h to 4j. Similarly, 4i could be interpreted by the loss of C(3)-hydroxyl unit from 4g, too.
In MS 3 analysis, ions 4d -4j were detected when 4c (m/z 558), 4d (m/z 540), 4e (m/z 526), and 4g (m/z 508) acted as precursor ions, in which the ion 4k at m/z 466 was caused by the loss of CO from 4h. The MS 3 experiment of the ion 4j at m/z 476 was carried out to yield fragments 4l -4n, in which 4l at m/z 458 arose from the scission of C(3)-hydroxyl group, 4m at m/z 444 originated from the elimination of C (18)-methoxyl unit, and 4n at m/z 354 derived from the loss of C(14)-benzoyl group.
Based on MS 4 experiment, the precursor ion 4h at m/z 494 eliminated C(8)hydroxyl moiety to produce the ion 4j at m/z 476. Precursor ion 4i at m/z 490 afforded ions 4l at m/z 458 and 4o at m/z 368 by losing C(6)-methoxyl or C(14)benzoyl group. From precursor ion 4m at m/z 444, ions 4p at m/z 426 and 4r at m/z 322 were easily explained by the eliminations of H 2 O at C(3) position and PhCOOH at C(14) position, and the ion 4q at m/z 394 derived from 4p by losing C (4)-tertiary carbon and C(13)-hydroxy groups consecutively.
According to MS 5 analysis, the ion 4j (m/z 476) was selected as precursor ion to give rise to the ions 4m at m/z 444, 4p at m/ z 426, and 4q at m/z 394. All of them had been explained in MS 3 and MS 4 experiments.
Ultimately, the ion 4q at m/z 394 was the only precursor ion in MS 6 analysis to afford the ion 4s at m/z 366 by losing one molecule of CO.
For benzoylmesaconine (4), ions with high intensity also arose from the losses of moieties at C(16), C(6), C(1) and C(8) positions such as 4d, 4e, 4g, 4h, and 4j whose abundance was more than 90%. It suggested that C(8) position was the active location at which moieties were easily to be lost. However, the position for proton leaving when C(8)-moiety was eliminated as neutral loss was disputed. Many articles tended to the positions at C (15) or C(7) [18,19]. In this paper, the position C(7) was proposed so that the fragments could lose its C(16)-methoxyl moiety to form a double bond between C(15) and C (16) positions and after that ketone -enol tautomerism occurred to form a carbonyl at C(15) position. In MS 2 analysis, the [M þ H -AcOH -MeOH -CO] þ (1d) ion from aconitine (1) was obtained that could demonstrate the above-mentioned conjecture. The high intensity ions 1b, 1d, 2b, 3b, 3c, and 4d were all come from the loss of C(8) or C(16)-moiety. The discussions indicated that C(8) was the most active position and the activity of C(16) is slightly inferior in MS n experiments, the eliminations of moieties at C(8) and C (16) positions were the originations to generate base peak ions.

2.2.2.
The losses of C(6)-methoxyl, C(1)methoxyl, C(3)-hydroxyl units, and C(14) substituent groups The fragmentation routes of losing C(6)methoxyl group after losing the C(8) and C (16) moieties were proposed in MS n analyses of the candidates. It was benefited from forming more stable ions with high conjugated structures such as 1e, 1f, 2e, 2f, 3e, 3f, 3i, 4e, 4g, and 4i, in which the ions 1f at m/z 494 and 4g at m/z 508 in MS 2 analysis were base peak ions simultaneously. After that, the eliminations of C(1)-methoxyl and C(3)-hydroxyl units occurred to yield fragments with longer conjugated structures. The loss of C(1)methoxyl moiety generated a series of ions as base peaks, for example, the ion 1i at m/ z 462 from aconitine (1), the ion 2i at m/z 334 from yunnaconitine (2), the ion 3g at m/z 488 from crassicauline A (3) and the ion 4j at m/z 476 from benzoylmesaconine (4) in MS 3 experiments. In addition, the ions originated from the elimination of C (14) substituent groups as PhCOOH or p-OMePhCOOH segment were also base peak ions, for instance, the ion 1l at m/z 354 from aconitine (1) in MS 3 analysis and the ion 3j at m/z 336 from crassicauline A (3) in MS 4 experiments. The discussions showed that the losses of C(6)-methoxyl, C(1)-methoxyl moieties and C(14) substituent groups were easily obtained and generated base peak ions readily following with C(8), C(15), and C(16) substituent groups.

The losses of N-ethyl/methyl
The loss of N-ethyl/methyl moiety was unfavorable on aconitine (1), yunnaconitine (2), and benzoylmesaconine (4). For crassicauline A (3), N-ethyl unit displayed as an active group to be eliminated easily which gave rise to the ions 3d at m/z 524 and 3r at m/z 276 in MS 3 experiments with high abundance. It could be explained that the absence of C(3)-hydroxyl influenced the activity of N-ethyl moiety and drove Nethyl group much more active in mass spectrometry. Meanwhile, the types of substituent groups at nitrogen atom impacted on the activity of moieties as well. For example, it was hard to detect the [P-CH 4 ] þ (P: product ion) ion in MS n analyses of benzoylmesaconine (4) with N-methyl moiety. It suggested that N-ethyl was more active than N-methyl in MS n analyses of C 19 -diterpenoid alkaloids. Thus, N-ethyl moiety of C 19 -diterpenoid alkaloids without C(3)-hydroxyl group was more active than the same moiety on the compounds with C(3)-hydroxyl unit or more active than N-methyl group.