Unusual Vilasinin-Class Limonoids from Trichilia rubescens

Eight vilasinin-class limonoids, including the unusually chlorinated rubescins K–M (1–3), the 2,3-epoxylated rubescin N (4), and rubescins O–R (5–8), were newly isolated from Trichilia rubescens. The structures of the isolated compounds were determined through spectroscopic and spectrometric analyses, as well as ECD calculations. The natural occurrence of chlorinated limonoids 1–3 was confirmed by chemical methods and HPLC analysis of a roughly fractionated portion of the plant extract. Eight selected limonoids, including previously known and new compounds, were evaluated for antiproliferative activity against five human tumor cell lines. All tested limonoids, except 8, exhibited significant potency, with IC50 values of <10 μM; in particular, limonoid 14 strongly inhibited tumor cell growth, with IC50 values of 0.54–2.06 μM against all tumor cell lines, including multi-drug-resistant cells.


Introduction
Tropical rainforests are renowned for their extraordinary biodiversity, which is the result of intricate ecosystems, a diverse array of species, and extensive genetic diversity within those species.As part of ongoing research focused on the phytochemical exploration of tropical rainforests, we investigated Trichilia rubescens (Meliaceae), which displayed significant antiproliferative activity in an NCI-60 panel screening (Figure S1 in the Supplementary Materials).

Results and Discussion
A 50% MeOH/CH2Cl2 extract (N047159) of the leaves of T. rubescens was partitioned with EtOAc and water.The EtOAc-soluble fraction was subjected to a series of chromatographic techniques using silica gel and octadecylsilica (ODS) gel via medium-pressure liquid chromatography (MPLC), column chromatography (CC), preparative TLC (pTLC), and HPLC to obtain pure new limonoids 1-8, along with the known compounds 9-14.
The structures of all known compounds were identified by a comparison of their 1D NMRs with previously reported values [6][7][8][9].
Compound 1 was isolated as a colorless amorphous solid with a specific rotation of []  21 −22 (c 0.1, CHCl3).The HRAPCIMS spectrum of 1 showed protonated molecular peaks at m/z 595.2118 and 597.2100 [M+H] + (calcd 595.2099 and 597.2069) in a 3:1 ratio, which suggested the presence of chlorine and a molecular formula of C33H35ClO8.The IR absorptions at 3257 and 1720 cm −1 implied the presence of hydroxy and carbonyl groups, respectively.The 1 H NMR spectrum showed signals assignable to protons for a monosubstituted phenyl (δH 7.92, 7.62, 7.47), three olefinic, three methylene, eight methine, and four methyl groups (Table 1).The 13 C NMR spectrum of 1 (Table 2) presented 33 peaks, including the signals for a ketone carbonyl (δC 199.4), an ester carbonyl (δC 165.9), four olefinic, and seven tertiary carbons.These signals were similar to those of TS2 (Figure S66 in the Supplementary Materials), a vilasinin-class limonoid with two β-epoxy groups at C-9, -11 and C-14, -15, a methacrylate at C-7, and an α,β-unsaturated ketone in ring-A, which was isolated previously from the leaves of T. rubescens [9].However, the chemical shifts of C-2, C-3, and the ester moiety at C-7 in 1 differed significantly from those in TS2; therefore, compound 1 likely lacks a C-2, C-3 double bond and has a different ester group at C-7.The planar structure of 1 was further confirmed by 2D NMR experiments (Figure 4).

Results and Discussion
A 50% MeOH/CH 2 Cl 2 extract (N047159) of the leaves of T. rubescens was partitioned with EtOAc and water.The EtOAc-soluble fraction was subjected to a series of chromatographic techniques using silica gel and octadecylsilica (ODS) gel via medium-pressure liquid chromatography (MPLC), column chromatography (CC), preparative TLC (pTLC), and HPLC to obtain pure new limonoids 1-8, along with the known compounds 9-14.The structures of all known compounds were identified by a comparison of their 1D NMRs with previously reported values [6][7][8][9].
Compound 1 was isolated as a colorless amorphous solid with a specific rotation of [α] 21  D −22 (c 0.1, CHCl 3 ).The HRAPCIMS spectrum of 1 showed protonated molecular peaks at m/z 595.2118 and 597.2100 [M+H] + (calcd 595.2099 and 597.2069) in a 3:1 ratio, which suggested the presence of chlorine and a molecular formula of C 33 H 35 ClO 8 .The IR absorptions at 3257 and 1720 cm −1 implied the presence of hydroxy and carbonyl groups, respectively.The 1 H NMR spectrum showed signals assignable to protons for a monosubstituted phenyl (δ H 7.92, 7.62, 7.47), three olefinic, three methylene, eight methine, and four methyl groups (Table 1).The 13 C NMR spectrum of 1 (Table 2) presented 33 peaks, including the signals for a ketone carbonyl (δ C 199.4), an ester carbonyl (δ C 165.9), four olefinic, and seven tertiary carbons.These signals were similar to those of TS2 (Figure S66 in the Supplementary Materials), a vilasinin-class limonoid with two β-epoxy groups at C-9, -11 and C-14, -15, a methacrylate at C-7, and an α,β-unsaturated ketone in ring-A, which was isolated previously from the leaves of T. rubescens [9].However, the chemical shifts of C-2, C-3, and the ester moiety at C-7 in 1 differed significantly from those in TS2; therefore, compound 1 likely lacks a C-2, C-3 double bond and has a different ester group at C-7.The planar structure of 1 was further confirmed by 2D NMR experiments (Figure 4). 1 H-1 H COSY correlations were observed between OH-3/H-3, H-2/H-3, H-5/H-6/H-7, H-11/H-12, H-16/H-17, and H-22/H-23.The HMBC correlations from H 3 -19 to C-1 and H-2 to C-1 suggested a carbonyl at C-1, while the correlations from H 3 -29 to C-3 indicated a hydroxy group at C-3.A benzoyl ester was assigned at C-7, based on the HMBC correlations of H-7 and H-3 ′ with C-1 ′ .The positions of the two epoxy groups at C-9, C-11 and C-14, C-15 were confirmed, based on multiple HMBC cross-peaks from H 3 -18, H 2 -16, and H 3 -30 to oxygenated carbons at C-14, C-15, and C-9, respectively.A pendant furan ring at C-17 was suggested by HMBC correlations from H-17 to C-20 and C-22, as well as from H-23 to C-20 and C-21.These data indicate that compound 1 has the same basic skeleton as TS2, but has a chlorine at C-2, a hydroxy at C-3, and a benzoate rather than methacrylate at C-7.The key NOESY correlations (Figure 5) from H-2 to H 3 -19/H 3 -29, H-6 to H-7/H 3 -19/H 3 -29, H-7 to H-15/H 3 -30, H-15 to H 3 -30, H-12α to H 3 -18/H-11, and H-12β to H-17 suggested α-orientations for the chlorine at C-2, the methyl at C-18, the ester carbonyl at C-7, and the furan ring at C-17; the latter three assignments were consistent with those of other related vilasinin-class limonoids, such as TS2 [9].The absolute configuration of 1 was deduced by comparing the experimental and calculated ECD spectra (Figure 6).Thus, the structure of 1 (rubescin K) was elucidated as the first chlorinated vilasinin-class limonoid, with (2S, 3R, 4R, 5S, 6R, 7S, 8S, 9S, 10S, 11S, 13S, 14R, 15R, 17S) absolute configurations.   1 and 2) and 2D (Figures 3 and 4) NMR spectra of compounds 1-3 strongly suggested that all three compounds are 2-chloro-3-hydroxy vilasinin limonoids.The only differences were found in the signals assignable to the C-7 ester moiety.The 13 C and 1 H NMR spectra of compound 2 suggested a tigloyl group, with an olefinic methine and two methyl carbons, as well as an olefinic and two sets of methyl protons assigned to the ester moiety.Additionally, an HMBC correlation between H-3 ′ and C-1 ′ supported the presence of a tigloyl ester, and a good comparison was found with the NMR data of the related tigloyl limonoid, compound 9 [6].The ester moiety of 3 was elucidated as a cinnamate, based on NMR observations of five phenyl methine and two olefinic protons, as well as HMBC correlations from olefinic protons H-2 ′ and H-3 ′ to phenyl carbons C-4 ′ and C-5 ′ .Although limited isolations of related cinnamoyl limonoids from the Meliaceae have been reported, the comparisons of the NMR data for the ester moiety of compound 3 with those of other cinnamoyl limonoids, such as toosendansins E and F [17], 7-cinnamoyltoosendanin [18], and ohchininolide [19] and its derivatives [20], also supported the presence of a cinnamate at C-7 in 3. Unfortunately, the instability of compounds 2 and 3 did not allow further investigation, such as with UV, IR, and ECD; however, all other data strongly supported that, like 1, both compounds are interesting chlorinated limonoids.Compound 4 was isolated as a colorless amorphous solid with an optical rotation of []  21 −13 (c 0.1, CHCl3).With a protonated molecular peak at m/z 455.2070 [M+H] + (calcd.455.2064) in its HRFABMS spectrum, the molecular formula of 4 is C26H30O7.The 1 H and 13 C NMR data (Tables 3 and 4) of 4 were like those of 10 (Figure 3) [7], a vilasinin-class limonoid with an oxetane ring formed by an ether linkage between C-7 and C-14.However, the signals that were due to the olefinic protons (H-2/H-3) in 10 were replaced by signals for protons (δH 3.15 and δH 7.57) on oxygenated carbons in 4, with a 1 H-1 H COSY correlation between the two protons.The 13 C NMR signals for C-2 and C-3 in 4 (δc 51.9 and δc 60.4) were shielded from those in 10 (δc 131.3 and δc 152.7) [7].The above data together with HMBC correlations between H-2 and C-4/C-10, H-3 and C-4/C-5, and H3-29 and C-3 indicated the presence of an epoxy ring at C-2/C-3.The relative configurations of 4 were established from a NOESY experiment (Figure 5).The β-orientation of the 2,3epoxy moiety was suggested by the key correlations of H-3/H2-28α and H2-28α/H-5.The NOESY correlations from H-7 to H-15/H3-30 and from H-15 to H3-30 also indicated a βorientation for OH-15.A strong NOE correlation was observed between β-oriented H-7 and α-oriented H-15; both protons are pseudo-equatorial, with a proton-proton distance of 2.4 Å.The same relative configurations were also supported by a comparison with all NMR data for compound 10 [7].The absolute configuration was supplied from the calculated and experimental ECD spectra (Figure 6).Thus, compound 4 (rubescin N) is (2R, 3R, 4R, 5S, 6R, 7S, 8S, 9S, 10S, 11S, 13S, 14S, 15R, 17S)-2,3-epoxyrubescin N, the first reported 2,3-epoxylated vilasinin-class limonoid.Furthermore, it is the second reported vilasinin- The 1 H and 13 C NMR data (Tables 3 and 4) of 4 were like those of 10 (Figure 3) [7], a vilasinin-class limonoid with an oxetane ring formed by an ether linkage between C-7 and C-14.However, the signals that were due to the olefinic protons (H-2/H-3) in 10 were replaced by signals for protons (δ H 3.15 and δ H 7.57) on oxygenated carbons in 4, with a 1 H-1 H COSY correlation between the two protons.The 13 C NMR signals for C-2 and C-3 in 4 (δc 51.9 and δc 60.4) were shielded from those in 10 (δc 131.3 and δc 152.7) [7].The above data together with HMBC correlations between H-2 and C-4/C-10, H-3 and C-4/C-5, and H3-29 and C-3 indicated the presence of an epoxy ring at C-2/C-3.The relative configurations of 4 were established from a NOESY experiment (Figure 5).The β-orientation of the 2,3-epoxy moiety was suggested by the key correlations of H-3/H 2 -28α and H 2 -28α/H-5.The NOESY correlations from H-7 to H-15/H 3 -30 and from H-15 to H 3 -30 also indicated a β-orientation for OH-15.A strong NOE correlation was observed between β-oriented H-7 and α-oriented H-15; both protons are pseudo-equatorial, with a proton-proton distance of 2.4 Å.The same relative configurations were also supported by a comparison with all NMR data for compound 10 [7].The absolute configuration was supplied from the calculated and experimental ECD spectra (Figure 6).Thus, compound 4 (rubescin N) is (2R, 3R, 4R, 5S, 6R, 7S, 8S, 9S, 10S, 11S, 13S, 14S, 15R, 17S)-2,3-epoxyrubescin N, the first reported 2,3-epoxylated vilasinin-class limonoid.Furthermore, it is the second reported vilasinin-class limonoid with an oxetane ring [7], although several other classes of oxetane limonoids have been isolated [21][22][23][24].As a biogenetic oxetane formation has been proposed previously [21,23], compound 4 might be produced from TS1 (Figure S66), which was also isolated from the same plant [9] through 2,3-epoxidation and the intramolecular nucleophilic attack of OH-7 on C-14 to open the epoxy ring.
Compound 7 (rubescin Q) has a molecular formula of C 26 H 28 O 7 , as suggested by HRFABMS.The 1D and 2D NMR spectroscopic data (Tables 3 and 4) of 7 indicated a comparable structure to that of 11 [7], in which a cyclopropane ring is formed between C-7, C-8, and C-14.However, the pendant furan ring found at C-17 in 11 was not present in 7. Instead, a 5-hydroxy-2-oxo-dihydrofuran ring was present in 7, based on the HMBC correlations (Figure 4) from H-17 to C-20 (δ C 138.8/138.9)and from H-23 to C-20 (δ C 138.8/138.9)and H-22 to C-21 (δ C 171.3/171.0),together with a 1 H-1 H COSY correlation between H-22 and H-23.Also, the related chemical shifts were identical to those of trichirubine A, which has the same substituent at C-17 [10].The 1:1 pairwise signals observed in the 1 H and 13 C NMR spectra indicated that 7 was a diastereomixture caused by the epimerization of a hemiacetal moiety at C-23.The same observation was also made with rubescin G [7] and trichirubine A [10].Based on the relative configurations revealed by NOESY correlations (Figure 4), the NOESY cross-peaks of H-7/H 3 -30/H-15 confirmed that a hydroxy group at C-15 and a proton at C-7 were β-oriented.The rigid conformation around the cyclopropane ring necessarily allowed the construction of the R configuration at C-14.The calculated and experimental ECD (Figure 6) elucidated the absolute configuration of 7 as (4R, 7S, 8S, 9S, 10S, 11S, 13S, 14R, 15R, 17R).
HRFABMS suggested a molecular formula of C 26 H 30 O 6 for compound 8 (rubescin R).A comparison of the NMR data (Tables 3 and 4) of 8 with those of compound 11 [7] (Figure 3) showed overall similarity, except at C-5, C-6, C-7, C-28, and C-29.The most significant differences were the shielded C-5 (δ c 55.4) and the deshielded C-6 (δc 207.1) and C-7 (δc 44.4) in 8 compared with 11, together with the appearance of a second hydroxy proton (δ H 1.64, 28-OH) and an aliphatic methine proton (δ H 3.67, H-5).These data suggested the absence of a C-5, C-6 double bond, the presence of a carbonyl at C-6, and the cleavage of the ether linkage between C-6 and C-28.The HMBC correlations of H-5 and H-7 with C-6 also supported the presence of a carbonyl at C-6 (δ C 207.1), while the shielded carbon signal for C-28 (δ C 70.3), the COSY connectivity of H 2 -28/OH-28, and the HMBC cross-peaks of H 2 -28 to C-4/C-5 indicated the presence of a hydroxymethyl moiety at C-4.While a NOESY analysis (Figure 5) confirmed the relative configuration of 8, an ECD analysis (Figure 6) and the negative specific rotation value determined that 8 and 11 have the same absolute configuration (4R, 5S, 7S, 8S, 9S, 10S, 11S, 13S, 14R, 15R, 17R).All the above data indicate that 8 is a seco-vilasinin-class limonoid and is a possible biosynthetic intermediate of 11.
A possible biosynthetic pathway to the chlorinated limonoids 1-3 could involve the epoxidation of related limonoids, such as 5, 9, and 15, which contain an α,β-unsaturated ketone in ring-A to give the tri-epoxy 16, followed by chlorination at C-2 (Figure 7).Limonoids 5 and 9 were isolated in this study.While halogenation is thought to primarily be a final biosynthetic step, limited evidence has indicated that halogenation can also be followed by subsequent steps that lead to other biosynthetic intermediates [25].Accordingly, there is a slight possibility that an epoxy group might be produced by the attack of a hydroxy group and the removal of a neighboring chlorine atom.
Molecules 2024, 29, x FOR PEER REVIEW 11 of 18 Limonoids 5 and 9 were isolated in this study.While halogenation is thought to primarily be a final biosynthetic step, limited evidence has indicated that halogenation can also be followed by subsequent steps that lead to other biosynthetic intermediates [25].Accordingly, there is a slight possibility that an epoxy group might be produced by the attack of a hydroxy group and the removal of a neighboring chlorine atom.Halogenated natural products are produced mainly by marine organisms living in halogen-rich environments, or by microorganisms such as algae, cyanobacteria, and fungus [26][27][28][29][30][31].Although rarely isolated from terrestrial plants, several plant-derived natural products containing halogen, usually chlorine, have been reported [32, [33][34][35].However, some of these compounds could be artifacts that are produced when halogenated solvents are used.Since this study is the first to report chlorinated limonoids and halogen-containing natural products from the family Meliaceae, further investigation was needed to determine whether they occur naturally in the plant extract.
Chlorination of the related β-2,3-epoxy limonoid 16 (Figure 7) could happen during the isolation process, for example, with the use of CHCl3 or CH2Cl2 under acidic conditions, such as on silica gel, or unexpectedly, by contamination with HCl.To investigate the possibility that the isolated chlorinated limonoids might be artifacts, the model substrate 17 was prepared as a biosynthetic precursor mimic of 1.The known limonoid 11, which was isolated in sufficient quantities in this study, was epoxidized via a general condition using H2O2 to give 17 at a 68% yield (Figure 8) [36].The β-orientation of the 2,3epoxide was confirmed by a NOESY correlation between H-3 and H-28α, which was also seen with 4. The treatment of 17 with excess silica gel in CHCl3 for 12 h produced no reaction; only the starting material was recovered.This observation suggested a low probability of chlorination occurring during silica gel column chromatography with a chlorinated solvent.The treatment of 17 with dried HCl on silica gel in CHCl3 gave a complex inseparable mixture rather than a chlorinated product [37].The 1 H-NMR spectrum of the mixture showed no characteristic peaks at H-2 and H-3 corresponding to the 2-chloro-3-hydroxy segment in 1.This result indicated that the chlorination of a related epoxide was unlikely, even after unexpected contamination with HCl.Halogenated natural products are produced mainly by marine organisms living in halogen-rich environments, or by microorganisms such as algae, cyanobacteria, and fungus [26][27][28][29][30][31].Although rarely isolated from terrestrial plants, several plant-derived natural products containing halogen, usually chlorine, have been reported [32][33][34][35].However, some of these compounds could be artifacts that are produced when halogenated solvents are used.Since this study is the first to report chlorinated limonoids and halogen-containing natural products from the family Meliaceae, further investigation was needed to determine whether they occur naturally in the plant extract.
Chlorination of the related β-2,3-epoxy limonoid 16 (Figure 7) could happen during the isolation process, for example, with the use of CHCl 3 or CH 2 Cl 2 under acidic conditions, such as on silica gel, or unexpectedly, by contamination with HCl.To investigate the possibility that the isolated chlorinated limonoids might be artifacts, the model substrate 17 was prepared as a biosynthetic precursor mimic of 1.The known limonoid 11, which was isolated in sufficient quantities in this study, was epoxidized via a general condition using H 2 O 2 to give 17 at a 68% yield (Figure 8) [36].The β-orientation of the 2,3-epoxide was confirmed by a NOESY correlation between H-3 and H-28α, which was also seen with 4. The treatment of 17 with excess silica gel in CHCl 3 for 12 h produced no reaction; only the starting material was recovered.This observation suggested a low probability of chlorination occurring during silica gel column chromatography with a chlorinated solvent.The treatment of 17 with dried HCl on silica gel in CHCl 3 gave a complex inseparable mixture rather than a chlorinated product [37].The 1 H-NMR spectrum of the mixture showed no characteristic peaks at H-2 and H-3 corresponding to the 2-chloro-3-hydroxy segment in 1.This result indicated that the chlorination of a related epoxide was unlikely, even after unexpected contamination with HCl.
Molecules 2024, 29, x FOR PEER REVIEW 11 of 18 Limonoids 5 and 9 were isolated in this study.While halogenation is thought to primarily be a final biosynthetic step, limited evidence has indicated that halogenation can also be followed by subsequent steps that lead to other biosynthetic intermediates [25].Accordingly, there is a slight possibility that an epoxy group might be produced by the attack of a hydroxy group and the removal of a neighboring chlorine atom.Halogenated natural products are produced mainly by marine organisms living in halogen-rich environments, or by microorganisms such as algae, cyanobacteria, and fungus [26][27][28][29][30][31].Although rarely isolated from terrestrial plants, several plant-derived natural products containing halogen, usually chlorine, have been reported [32,[33][34][35].However, some of these compounds could be artifacts that are produced when halogenated solvents are used.Since this study is the first to report chlorinated limonoids and halogen-containing natural products from the family Meliaceae, further investigation was needed to determine whether they occur naturally in the plant extract.
Chlorination of the related β-2,3-epoxy limonoid 16 (Figure 7) could happen during the isolation process, for example, with the use of CHCl3 or CH2Cl2 under acidic conditions, such as on silica gel, or unexpectedly, by contamination with HCl.To investigate the possibility that the isolated chlorinated limonoids might be artifacts, the model substrate 17 was prepared as a biosynthetic precursor mimic of 1.The known limonoid 11, which was isolated in sufficient quantities in this study, was epoxidized via a general condition using H2O2 to give 17 at a 68% yield (Figure 8) [36].The β-orientation of the 2,3epoxide was confirmed by a NOESY correlation between H-3 and H-28α, which was also seen with 4. The treatment of 17 with excess silica gel in CHCl3 for 12 h produced no reaction; only the starting material was recovered.This observation suggested a low probability of chlorination occurring during silica gel column chromatography with a chlorinated solvent.The treatment of 17 with dried HCl on silica gel in CHCl3 gave a complex inseparable mixture rather than a chlorinated product [37].The 1 H-NMR spectrum of the mixture showed no characteristic peaks at H-2 and H-3 corresponding to the 2-chloro-3-hydroxy segment in 1.This result indicated that the chlorination of a related epoxide was unlikely, even after unexpected contamination with HCl.To further confirm that the isolated chlorinated limonoids are natural products, LC/MS analysis was carried out on the initial fraction 7b containing 1, which was roughly twice separated from the original extract using silica gel CC.The peak with the exact mass for an authentic sample of 1 was observed within a range of 4.3-4.5 min, and the same peak was detected in fraction 7b at the same retention time (Figures S64 and S65 in the Supplementary Materials).
Based on the results of the above chemical methods and LC/MS analysis, it is highly possible that chlorinated limonoids 1-3 occur naturally in T. rubescens.
Eight selected vilasinin-class limonoids (1, 5, 6, 8, 9, 11, 13, and 14) were evaluated for their antiproliferative activities against human tumor lines (HTCLs), A549 (lung adenocarcinoma), MDA-MB-231 (triple-negative breast cancer), MCF-7 (HER2-negative), KB (HeLa derivative), and KB-VIN [P-gp overexpressing multidrug-resistant (MDR) KB subline] (Table 5).All the tested limonoids, except 8, showed significant activity, with IC 50 values of 0.54-8.46µM, even against the KB-VIN MDR cell line, which suggested that these limonoids are not P-gp substrates.In particular, compound 14 exhibited the highest potency against all HTCLs, including the MDR tumor cell line.Compounds 9 and 14 have also been reported to effectively decrease the viability of hepatoma cells at TC 50 concentrations ranging from 5 to 10 µM.[38].Comparing the results of the current study, these two compounds show potential value as anticancer drugs.The chlorine-containing limonoid 1 showed slightly selective inhibition against the MCF-7 cell line (IC 50 1.34 µM) compared with the other tested HTCLs (IC 50 2.53-5.72 µM).Compounds 1 and 5 showed similar activity, indicating that the 2,3-double bond is not very important for this activity.The comparison of 5, 9, and 14 suggested that the function group at C-7 has an insignificant effect on this activity, although compound 14, with a 6,7-double bond, displayed slightly more potent activity than other compounds.Since compound 8 did not exhibit significant antiproliferative activity (IC 50 > 40 µM), the tetrahydrofuran ring, which is formed by an ether linkage between C-6 and C-28, might be important for the antiproliferative activity of this compound type [39].

Plant Material
A 50% CH 2 Cl 2 /MeOH extract of T. rubescens leaves (N047159) was provided by the NCI Natural Products Branch (Developmental Therapeutics Branch, Frederick, MD, USA) as reported previously [40].The plants were collected in the Central Africa Republic by J. M. Fay in August 1988 and were identified by the taxonomist R. Gereau.

Rubescin
Preparation of compound 17: 5% NaOH (0.2 mL) and 30% H 2 O 2 (0.099 mL) were added to compound 11 (17.0 mg) in MeOH (1.0 mL) and the solution was stirred at room temperature for 11.5 h.The completion of the reaction was confirmed with TLC.The reaction mixture was quenched with the 1M HCl (20 mL).The resulting mixture was then extracted 3 times with EtOAc.The organic layer was dried over Na 2 SO 4 , filtered, and evaporated under vacuum [36].The crude product was purified via silica gel CC (n-hexane: EtOAc; 1:1) to yield compound 17 as a colorless amorphous solid (12.0 mg).

Calculation of ECD Spectra
Preliminary conformational analyses of all compounds, except compound 4, were performed by CONFLEX9 with the MMFF94 force field.Spaltan20 was used for the preliminary conformational analysis of compound 4. The obtained conformers were further optimized in MeOH by the density functional theory (DFT) method, with the B3LYP functional and 6-31(d) basis set.The ECD spectrum was calculated via the time-dependent DFT (TDDFT) method, using the CAM-B3LYP functional and TZVP basis set.The calculation was performed using the conformers within 2 kcal/mol predicted in MeOH.The solvent effect was introduced by the conductor-like polarizable continuum model (CPCM).The DFT optimization and TDDFT-ECD calculation were accomplished by Gaussian16 (Gaussian, Inc., Wallingford, CT, USA).The calculated spectrum was displayed using GaussView 6.1, with the peak half-width at half-height being 0.333 eV.The Boltzmann-averaged spectrum at 298.15 K was calculated using Excel 2016 (Microsoft Co., Redmond, WA, USA).The calculations were re-optimized according to the literature [41].

Antiproliferative Activity Assay
A549, KB, MDA-MB-231 and MCF-7 were obtained from the Lineberger comprehensive Cancer Center (UNC-CH, NC) or from ATCC (Manassas, VA, USA).KB-VIN was a generous gift of Professor Y.-C.Cheng (Yale University, New Haven, CT, USA).We confirmed our KB and KB-VIN are identical to AV-3 (ATCC number, CCL-21) as a HeLa (cervical carcinoma) derivative by short tandem repeat (STR) profiling.The antiproliferative activity was examined via an SRB assay, as described previously [42].Briefly, freshly trypsinized cell suspensions were seeded in 96-well microtiter plates at densities of 4000-11,000 cells per well for each compound.The attached cells were fixed in 10% trichloroacetic acid and stained with 0.04% SRB after culturing for 72 h.The absorbance at 515 nm was measured using a microplate reader (Spark 10M, Tecan, Zurich, Switzerland) with SparkControl software version 2.3 (Tecan) after solubilizing the bound dye with 10 mM Tris base.The mean IC 50 was determined as the average from at least three independent experiments of duplication for an assay and similar determinations.

Conclusions
The phytochemical investigation of a 50% MeOH/CH 2 Cl 2 extract from the leaves of a tropical rainforest plant, T. rubescens, led to the isolation of 14 vilasinin-class limonoids, including the new rubescins K-R (1-8) and known rubescins E, F, and H-J (9-13), as well as TS3 ( 14), together with three sesquiterpenes and two sterols.An extensive analysis of isolated compounds revealed that rubescins K-M (1-3) were unusually chlorinated limonoids and rubescin N (4) was the first 2,3-epoxylated vilasinin-class limonoid.The natural occurrence of chlorinated limonoids was further confirmed using chemical methods and LC/MS analysis.
The isolated vilasinin-type limonoids 1, 5, 6, 8, 9, 11, 13, and 14 were evaluated for their growth-inhibitory effects against five human tumor cell lines, including a multidrugresistant cell line, KB-VIN.All the tested limonoids, except for compound 8, exhibited significant activity, with IC 50 values of 0.54-8.46µM against all tested tumor cell lines, including KB-VIN.Compound 14 showed the highest inhibitory activity, while chlorinated limonoid 1 demonstrated slightly selective inhibition against the MCF-7 cell line.The preliminary structure-activity relationship study indicated that the tetrahydrofuran ring formed by C4-6 and C-28, which is characteristic of the vilasinin-class limonoid, might be important for this activity.

Figure 8 .Figure 7 .
Figure 8. Investigation of the possibility of an artifact using a model substrate 11.

Figure 8 .Figure 8 .
Figure 8. Investigation of the possibility of an artifact using a model substrate 11.
a Broad singlet.

Table 5 .
Antiproliferative activity of the selected limonoids.