Quassinoids represent a class of nortriterpene compounds, which are commonly recognized as characteristic metabolites of the Simaroubaceae family [
1–
5]. Structurally, these compounds typically feature a pentacyclic or tetracyclic lactone core and are categorized into six subtypes: C
26, C
25, C
20, C
19, C
18, and C
22 types [
6]. To date, approximately 480 naturally occurring quassinoids have been identified, exhibiting a wide range of significant biological activities, such as antifeedant, antitumor, anti-inflammatory, and antimalarial effects [
5,
7–
11]. As a prominent member of the Simaroubaceae family,
Brucea javanica has garnered considerable research interest in recent decades for a variety of biologically active quassinoids [
12–
16]. Using advanced technological tools to help us rapidly and accurately identify quassinoids with novel skeletons in
B. javanica, and expand the chemical diversity of quassinoids should be our primary focus.
Interdisciplinary studies crossing the natural sciences and computer science make it possible to effectively discover natural products with novel skeletons by utilizing the advanced technology [
17]. Based on the rapid advance of information technology, nuclear magnetic resonance spectroscopy (NMR)-based analytical techniques for complex chemical mixtures have substantially accelerated the identification of new natural products [
18–
25]. In parallel, tandem mass spectrometry-based molecular networking has also evolved to become a frequently used method. Such tools were indispensable for contemporary phytochemical research to efficiently identify structurally novel natural products with low natural abundance [
26–
31]. These technological advancements revolutionize natural product discovery, but their comparative advantages and limitations have gradually emerged in practical applications.
Considering the characteristics of every single strategy, we advance a new integrated analytical strategy based on DeepSAT and molecular networking. DeepSAT accurately extracts the chemical features to achieve structure annotation and scaffold prediction of major constituents but a relatively lower sensitivity in trace compound detection [
18]. With MS/MS-based technology, its sensitivity dramatically increased over the DeepSAT method, and it precise localizes target compounds through molecular network mapping. Applying this approach, two novel hemiterpene-quassinoid adducts (bruquassins A and B,
1 and
2) with an unusual 5/6/6/6/6/5 hexacyclic scaffold were successfully identified from
B. javanicaas (
Fig. 1A). Encouragingly, these two compounds are the first reported examples of hemiterpene-quassinoid adducts. Herein, the guided isolation, structure elucidation, and hypothetical biogenetic pathways of
1 and
2 were described. Furthermore, we accomplished the biomimetic synthesis of
1 and evaluated the antifeedant activities of all these isolates against
Plutella xylostella (DBM, the abbreviation of diamondback moth).
The residual alcohol extract of
B. javanica was suspended in H
2O, extracted with
n-BuOH, and subsequently subjected to multiple chromatography, resulting in the isolation of four fractions (Fr. B1 to B4). The heteronuclear single quantum correlation (HSQC) spectrum and liquid chromatography-tandem mass spectrometry (LC-MS/MS) data of Fr. B1 to B4 from
B. javanica were analyzed by the integrated analytical strategy to prioritize the identification of intriguing quassinoids. The HSQC data were formatted into a comma-separated values (CSV) file and subsequently uploaded to the DeepSAT platform (https://deepsat.ucsd.edu/) [
18,
23]. As shown in
Fig. 1B, Fr. B4 has a 96.84% probability of being rich in triterpenoids, and the top six structures ranked by cosine similarity scores were all identified as quassinoids. It is worth noting that the low cosine similarity scores observed in the DeepSAT results suggested the presence of novel quassinoids skeletons in Fr. B4. Nevertheless, it is difficult to determine the location of potential quassinoids in Fr. B4 only by using DeepSAT. Therefore, to rapidly discover the possible novel quassinoids with low abundance, the GNPS workflow was chosen and applied [
27,
28,
32,
33]. In the generated molecular network (
Fig. 1C), the main ingredient in
B. javanica, bruceine A was initially identified [
34]. Notably, the molecular network contained several nodes with molecular weights exceeding 600, approximately 100 Da higher than those of reported quassinoids [
6], suggesting that these nodes were possible to possess complex quassinoids structures. Accordingly, guided by DeepSAT and molecular networking, two quassinoids with unparalleled skeletons were successfully isolated and characterized (
Fig. 1A).
Bruquass A (
1) was obtained as white amorphous powder with a chemical formula of C
31H
40O
12 based on its HRESIMS ion peak at
m/
z 605.2590 [M + H]
+ (calcd. for C
31H
41O
12, 605.2593), accounting for twelve degrees of unsaturation. The 1D NMR (Table S1 in Supporting information) together with HSQC spectrum of
1 revealed resonances for six methyl groups, including one methoxy, four singlets, and one doublets; four sp
3 methylene, including one oxygenated; eight sp
3 methine, including four oxygenated; five sp
3 quaternary carbons, including three oxygenated; four olefinic; an aldehyde group; and three ester groups. Based on the double bond equivalents, compound
1 possessed a six-ring framework. The
1H-
1H correlation spectroscopy (COSY) correlations of H
3-18/H-4/H-5/H
2-6/H-7, H-9/H-11/H-12, H-14/H-15, and the heteronuclear multiple bond correlation (HMBC) correlations from H
3-18 to C-3, H
3-19 to C-1/5/9/10, H
2-17 to C-7/8/9, H-7 to C-14, H-11 to C-13, and H-12 to C-20 manifested the presence of quassinoid motif (
Fig. 2A). Meanwhile, the existence of ring F and its fusion to ring A
via C-2 and C-3 were inferred from the HMBC correlations of 2-OH to C-1/2/1′′, H
3-3′′ to C-1′′/2′′/4′′, and H-5′′ to C-2/4′′. The assignment of the angeloyl group was done according to the double bond equivalents and the HMBC correlations from H
3-4′/5′ to C-2′/3′. Hence, the planar structure of
1 was unequivocally elucidated.
The relative configuration of
1 was resolved
via the nuclear Overhauser effect spectroscopy (NOESY) correlations. Based on the NOESY correlations (
Fig. 2B) between H-5 and H-9, H-9 and H-15, H-9 and H-11, 2-OH and 3-OH, 2-OH and H
3-19, H
3-19 and H-4, H
3-19 and H
2-17, H
2-17 and H-12, H
2-17 and H-14, H
2-17 and H-7, the relative configuration was partially assigned as 2
R*,3
S*,4
S*,5
S*,7
R*,8
R*,9
R*,10
S*,11
R*,12
S*,13
S*,14
S*,15
R*. The absolute configurations of
1 was confirmed by time-dependent density functional theory (TDDFT)/ECD calculations [
4,
32,
35–
38]. Finally, The ECD curves generated for (2
R,3
S,4
S,5
S,7
R,8
R,9
R,10
S,11
R,12
S,13
S,14
S,15
R)-
1 agreed well with the experimental curves for
1, allowing the assignment of the absolute configuration of
1 as 2
R,3
S,4
S,5
S,7
R,8
R,9
R,10
S,11
R,12
S,13
S,14
S,15
R (
Fig. 2C). Conclusively, by the systematic analysis above, we could unquestionably determine the compound
1 as an unparalleled adduct of hemiterpene and quassinoid with 5/6/6/6/6/5 hexyclic ring system.
Bruquass B (
2) was determined to have a molecular formula of C
31H
42O
12, 2Da more than that of
1, according to its protonated molecular ion at
m/
z 607.2745 in its HRESIMS data (calcd. for C
31H
43O
12, 607.2749). Comparison of the 1D and 2D NMR spectra of
2 with those of
1 revealed their similarity, expected that the angeloyl group in
1 was replaced by an isovaleric group (
Fig. 2A). The relative configurations of
2 was further ascertained by NOESY experiment (
Fig. 2B). Furthermore, the absolution configuration of
2 was ascertained to be 2
R,3
S,4
S,5
S,7
R,8
R,9
R,10
S,11
R,12
S,13
S,14
S,15
R according to the matched ECD curves between the calculated and experiment one (
Fig. 2C).
The biosynthetic pathway of
1 can plausibly be traced back to brusatol (
3), existing in
B. javanica at some level of abundant. As illustrated in
Scheme 1A, intermediate
Ⅰ was synthesized from brusatol (
3) and 3-methyl-2-butenal (
5) by aldol condensation [
39]. Owing to the inherent instability of the enol structure, intermediate
Ⅰ rapidly underwent tautomerization to its keto form (intermediate
Ⅱ). Subsequently, the elimination of the
γ-hydrogen atom generates a carbanion, with the resulting negative charge delocalized to the
α-position [
40]. Ultimately, compound
1 was formed
via intramolecular Aldol condensation. Similarly, compound
2 was proposed to be biosynthesized through an analogous pathway, with bruceine A (
4) serving as the precursor.
To provide a sufficient sample of
1 for biological evaluation and further accurate structural validation, a concise and viable biomimetic synthesis starting from brusatol (
3) was designed based on the hypothetic biogenetic pathway of
1 (
Scheme 1B). Considering that the formation of
γ-carbanion was critical in synthesis of
1, we attempted to catalyze the reaction between
3 and 3-methyl-2-butenal (
5) under different conditions [
41–
45]. the expected product
1 was not observed when acidic catalysts were employed (Table S2 in Supporting information, entries 1–4). By comparison, under basic catalytic conditions, the desired product
1 was successfully formed using sodium hydroxide or potassium hydroxide (Table S2, entries 5–8). Due to the presence of multiple ester bonds in the structure of
3, which are susceptible to decomposition under strong alkaline conditions, the subsequent reactions were carried out at 0 ℃ and the product
1 was formed in 16.9% yield (Table S2, entry 9). What is more, among the solvents screened, DMF proved to be the most effective (25.3% yield) for the reaction (Table S2, entry 10). In conclusion, the bionic synthesis of
1 was achieved by alkali-catalyzed vinylogous aldol condensation.
Guided by the biomimetic synthesis, sufficient compounds were prepared and purified to evaluate their antitumor activities. However, compounds
1 and
2 showed no cytotoxicity in HCT116, Caco-2, and HT-29 cell lines. In view of the antifeedant activity of quassinoids in
B. javanica [
16,
46,
47], the antifeedant activities against DBM of compounds
1 and
2 was screened by leaf disc method. The bioassay revealed significant concentration dependent antifeedant activity, with compound
1 exhibiting particularly potent effects (the concentration corresponding to 50% antifeedant action (AFC
50) = 26.58 µg/mL). Following incubation with seeding of
Brassica oleracea L. (Brassicaceae), compound
1 decreased the feeding of the DBM larvae on the leaves (
Figs. 3A and
C). After a 5-day culture period, the antifeedant index (AFI) remained above 70% after treatment with a high concentration (50 µg/mL) of compound
1 (
Fig. 3B). In addition, the body size of the DBM larvae fed with compound
1 was significantly smaller and lighter than that of the control group (
Fig. 3D).
In summary, we employed an integrated approach combining DeepSAT with tandem mass spectrometry-based molecular networking to search for novel bioactive quassinoid analogues from B. javanica. Guided by this approach, bruquass A and B (1 and 2), possessing unusual 5/6/6/6/6/5 hexyclic ring system, were identified as the first examples of quassinoid-hemiterpene hybrids. Their structure and absolute configurations were elucidated by comprehensive analysis of spectroscopic data and quantum chemical calculations. Notably, compound 1 was successfully synthesized and the corresponding optimal synthetic conditions were explored, processing sufficient compounds for biological experiments. Results of bioactivity study showed that the isolates exerted potent antifeedant activity on the Plutella xylostella. The discovery of these novel skeletal compounds not only enriches the structural diversity of quassinoids, but also furnishes promising candidates for the development of natural insecticides.
PDF
Full
58
33
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
