Structural Characterization, and Antioxidative and Anti-inflammatory Activities of Phylloxanthobilins in Tropaeolum majus , a Plant with Relevance in Phytomedicine

Abstract

Tropaeolum majus (garden nasturtium) is a plant with relevance in phytomedicine, appreciated not only for its pharmaceutical activities, but also for its beautiful leaves and flowers. Here, we investigated the phytochemical composition of senescent nasturtium leaves. Indeed, we identified yellow chlorophyll catabolites, also termed phylloxanthobilins, which we show to contribute to the bright yellow color of the leaves in the autumn season. Moreover, we isolated and characterized the phylloxanthobilins from T. majus , and report the identification of a pyro-phylloxanthobilin, so far only accessible by chemical synthesis. We show that the phylloxanthobilins contribute to bioactivities of T. majus by displaying strong anti-oxidative effects in vitro and in cellulo , and anti-inflammatory effects as assessed by COX-1 and COX-2 enzyme inhibition, similar to other bioactive ingredients of T. majus , isoquercitrin, and chlorogenic acid. Hence, phylloxanthobilins could play a role in the efficacy of T. majus in the treatment of urinary tract infections, an established indication of T. majus. With the results shown in this study, we aid in the completion of the phytochemical profile of T. majus by identifying additional bioactive natural products as relevant components of this medicinal plant.

Abbreviations

CGA

chlorogenic acid

COX

cyclooxygenase

CP

chemically pure

FRAP

ferric reducing antioxidant power

IQ

isoquercitrin

LPS

lipopolysaccharide

NCC

nonfluorescent chlorophyll catabolite

PB

phyllobilin

PleB

phylloleucobilin

PxB

phylloxanthobilin

pyPxB

pyro-phylloxanthobilin

ROS

reactive oxygen species

RT

room temperature

TPTZ

2,4,6-tri(2-pyridyl)-s-triazine

YCC

yellow chlorophyll catabolite  

Introduction

Lately, a chlorophyll-derived class of substances, PBs, has emerged as relevant among phytochemicals. PBs, known to accumulate in senescent leaves of vascular plants and peels of fruit, have not been fully investigated yet ¹ . Possessing tetrapyrrolic structures, they share structural similarities with bilirubin ² . In some of the investigated plants, chlorophyll degradation leads to the formation of PxBs [also termed yellow chlorophyll catabolites (YCCs)] ³ . In terms of structure, PxBs are oxidation products of earlier discovered PleBs (or nonfluorescent chlorophyll catabolites (NCCs)] ¹ , from which PxBs are formed in the plant cell by a yet unclarified mechanism ( Fig. 1 ) ⁴ . A structural feature of naturally occurring PleBs is a carboxyl group or methyl ester function at ring E. The loss of the carboxyl function has not been discovered in plants yet; in ferns, however, so-termed iso-phyllobilin s feature a similar structural motif from a rearrangement reaction ⁵ . Semisynthetic approaches, however, achieved the formation of so-called pyro-PleBs (pyro-NCCs) by decarboxylation of natural PleBs using pig liver esterase and aqueous 20 mM H ₂ SO ₄ at 80 °C for 6 h ⁶ . Contrary to the decarboxylation of PleBs, earlier intermediates in the breakdown of chlorophyll can easily decarboxylate; the reason for the stability of the carboxyl function of PleBs and PxBs in plants has not yet been revealed ⁵ ,  ⁶ .

Fig. 1.

 PleBs are oxidized to PxBs in plant vacuoles by a yet undeciphered mechanism.

Research in recent years has found that the breakdown of chlorophyll proceeds via a multistep biochemical pathway. Interestingly, the substances are formed in plants not only in autumn but also, e.g., in the event of a pathogen attack ¹ ,  ⁷ . This suggests that PBs could display plant protecting properties, as is known for other phytochemicals, such as flavonoids or saponins ⁸ ,  ⁹ ,  ¹⁰ . Data about chlorophyll, its derivatives, bioactivities, bioavailability, and antioxidative power is comprehensive ¹¹ . For example, chlorophyllin, a sodium-copper salt analogue of chlorophyll, protects mitochondria against oxidative damage caused by reactive oxygen species (ROS), and membranes in vitro and ex vivo by inhibiting the generation of ROS ¹² .

Regarding the products of the biochemical degradation of chlorophyll in the course of senescence, however, knowledge has only begun to accumulate recently. In order to render the identification of PBs in medicinal plants and food more straightforward, a PB spectral database has recently been published ¹³ . Current findings have shown that PxBs contained in food and medicinal plants possess strong antioxidative activities. In Echinacea purpurea , a plant commonly used to treat the common cold, six PxBs were found with strong in vitro and in cellulo antioxidative activity ¹⁴ . Furthermore, a PxB from stinging nettle leaves was found to possess strong antioxidant power, considerably contributing to the plantʼs spectrum of activity ¹⁵ . Oxidative stress occurs as result of an imbalance between the generation and cleavage of radical oxygen species during various physiological processes, and is pathologically related to numerous diseases, such as atherosclerosis, diabetes, and Parkinsonʼs disease ¹⁶ ,  ¹⁷ ,  ¹⁸ . Still, antioxidants are considered a promising tool to effectively conquer diseases ¹⁹ . Moreover, recent findings have pointed towards a suppressive potential of PBs on cellular immune activation pathways by inhibiting tryptophan catabolism to kynurenine in peripheral blood mononuclear cells (PBMCs), opening a complete view of the physiological effects of PBs ²⁰ .

Along these lines, the full characterization of the chemical components of Tropaeolum majus L. (garden nasturtium, Tropaeolaceae) remains incomplete, and the composition not only varies among different plant parts but is also influenced by cultivation methods and environmental conditions. The flowers represent the most analyzed part of the plant, and the phytochemical composition was found to be different depending on the color of the flowers ²¹ . In contrast, a limited number of studies have been conducted on nasturtium leaves, describing flavonoids, such as isoquercitrin, quercetin 3-glucoside, kaempferol, fatty acids, and glucosinolates, such as glucotropaeolin and sinalbin ²¹ ,  ²² ,  ²³ ,  ²⁴ . Garden nasturtium is widely cultivated and used as decorative food in, e.g., salads. In addition to the use as a food, nasturtium has also gained recognition in folk medicine for its diuretic and antibacterial effects. In Germany, T. majus is listed in the monographies of the commission E Nr. 162 of 1992 as phytomedicine in the treatment of urinary tract infections and catarrhs of the upper respiratory tract. Moreover, extracts of nasturtium leaves were reported by Gasparotto et al. to possess natriuretic and diuretic activity in vivo , and further, IQ is attributed as a bioactive key player of the multicomponent mixture ²⁵ . Polyphenols, mainly CGA and ascorbic acid, were concluded to be the antioxidative and anti-inflammatory key players in herbal extracts of nasturtium ²⁶ .

In this study, we report the identification of four PxBs in leaves of T. majus that give rise to the yellow color of leaves during senescence and possess strong antioxidant and anti-inflammatory potency comparable to CGA and IQ; hence, PxBs from nasturtium leaves may contribute to the effects of nasturtium as a medicinal plant. In addition, we uncover the first natural source of a pyPxB, which opens the door for investigating the bioactivities of an interesting structural motif.

Results and Discussion

Investigating the constituent fingerprint of aqueous methanolic extracts of T. majus leaves by analytical HPLC, four different PxBs were identified by their characteristic UV spectrum ( Fig. 2 ) ¹³ . As widely established for PBs, PxBs were named by the botanical source (prefix) and numbered Tm -PxB-(1 – 4) according to decreasing polarity. The characteristic UV/Vis spectrum of PxB has an absorbance maximum at 426 nm, which is due to rings A to D possessing an extended π-electron system and carrying a vinyl residue at ring D ³ . Tm -PxB-4, being the less polar, was detected to have maxima at 430, 314, and 246 nm. Tm -PxB-(1 – 3) showed a shifted maximum to a lower wavelength at 418 nm, besides the maxima at 314 – 316 nm and 242 – 246 nm, which was reported earlier for PxBs of which the vinyl group is replaced by a dihydroxyethyl residue at ring D in Egeria densa and E. purpurea ( Fig. 2S , Supporting Information) ¹⁴ ,  ²⁷ .

Fig. 2.

 Analytical HPLC trace of acidified methanolic extracts of senescent T. majus leaves, detection at 420 nm, long HPLC method: mobile phase A = ammonium acetate buffer 10 mM pH 7, B = ACN, flow 0.5 mL/min. Solvent composition: 0 – 5 min 5% B, 5 – 45 min 5% to 60%, 45 – 51 min 60% to 100%, 51 – 53 min 100%, 53 – 55 min 100% to 5%.

Relative content of PxBs in methanolic extracts of fresh yellow leaves to the sum of all peak areas in the 420 nm HPLC trace yielded a total of approximately 25% ( Figs. 3S and 4S , Supporting Information). Since the yellow color is widely associated with the wavelength at 420 nm, it can be concluded that PxBs are responsible for an estimated 25% of the yellow coloration of nasturtium leaves in autumn.

For the structural elucidation of PBs, a total of 200 g of senescent leaves were used for extraction, as described in the “General experimental procedures” section. The yield amounted to 12.3 mg of Tm -PxB-1, 0.8 mg of Tm -PxB-2, 1.1 mg of Tm -PxB-3, and 2.3 mg of Tm -PxB-4, as quantified by UV/Vis spectroscopy. HR-ESI-MS was performed to identify the elemental composition, and MS ² experiments were performed to elucidate side chain modifications via the characteristic fragmentation reactions according to Müller et al. ²⁸ . The molecular formulae were derived from the [M – H] ⁻ as well as [M + H] ⁺ molecular ions to be C ₃₄ H ₃₈  N ₄ O ₁₀ for Tm -PxB-1, C ₃₃ H ₃₈  N ₄ O ₈ for Tm -PxB-2, C ₃₄ H ₃₉  N ₄ O ₉ for Tm -PxB-3, and C ₃₄ H ₃₆  N ₄ O ₈ for Tm -PxB-4. Further MS ² experiments in the positive ion mode revealed characteristic fragmentations for all four PxBs, namely, the loss of H ₂ O, CO ₂ , as well as cleavages at the saturated meso positions between rings A and B as well as rings B and C ( Figs. 6S-9S , Supporting Information). Additionally, matching of databases indicated that Tm -PxB-1 may be the corresponding oxidation product of So -PleB-1 ( So -NCC-1) of spinach, also known as Mc -PleB-26 ( Mc -NCC-26), the most polar PleB found in ripening peels of banana ²⁹ ,  ³⁰ ,  ³¹ . To further verify the proposed structure, fresh spinach extracts were analyzed by HPLC. HPLC analysis revealed identical retention times of So -PleB-1 and the corresponding PleB in the nasturtium extract, indicating that a catabolite identical to So -PleB-1 is present in nasturtium leaves ( Fig. 5S , Supporting Information).

In addition to MS ² fragmentations that already provided information on the molecular structures ( Fig. 3 ), ¹ H-NMR spectra as well as homo- and heteronuclear 2D spectra (COSY, ¹ H, ¹³ C-HMQC, ¹ H, ¹³ C-HMBC) were acquired for Tm -PxB-1 and Tm -PxB-4 for further confirmation ( Figs. S11 – S15 , Supporting Information). The ¹ H-NMR spectra of Tm -PxB-1 and Tm -PxB-4 showed a set of typical signals for PBs. Among them were four characteristic signals around 2 ppm assigned to the methyl groups, as well as a characteristic signal at 9.5 ppm originating from the formyl group. The double bond between C15 and C16, a decisive factor for the yellow color of PxBs, was displayed by a signal at 5.96 ppm ( Tm -PxB-1) and 6.04 ppm ( Tm -PxB-4), respectively. While the spectrum of Tm -PxB-4 showed characteristic signals for a vinyl group at 5.26/6.32 and 6.50 ppm, these signals were replaced by high shifted signals, characteristic for the dihydroxy side chain of Tm -PxB-1 ( Table 1S , Supporting Information). For Tm -PxB-2, ¹ H-NMR spectra were recorded ( Fig. 13S , Supporting Information). Although only some of the characteristic signals could be assigned, a typical pattern for PxBs became evident; the four methyl groups showed singlets at 2.16, 1.98, 2.06, and 2.15 ppm. The protons of the dihydroxy side chain at C18 showed signals at 4.47 ppm and 3.46/3.68 ppm, respectively. Signals from pyrrole nitrogen-bound protons were visible at around 10 ppm, and the proton at C15 manifested in a signal at 5.98 ppm, supporting the presence of a double bond. In summary, results of ¹ H-NMR as well as tandem MS analyses strengthened Tm -PxB-2 to be a pyPxB-type chlorophyll degradation product ( Fig. 3 ).

Fig. 3.

 Chemical structure of the Tm -PxB core and side chain modifications.

To test whether the newly discovered pyPxB is formed during the extraction process and therefore is a possible degradation product of Tm -PxB-1, the stability of this PxB in different media that were used during extraction was tested by HPLC analysis. Indeed, Tm -PxB-1 was stable for 1 h of incubation at RT and still accounted for 100% in PBS pH 7 and PBS 5.2. Samples acidified with AcOH (pH 3.5), when left at RT overnight, showed a reduction of the Tm -PxB-1 peak area of 22% and 4% formation of Tm -PxB-2, 0.3% of Tm -PxB-3, and 1.6% of Tm -PxB-4. The behavior of Tm -PxB-1 in acidic PBS pH 2.5 overnight was additionally analyzed by LC-MS. The peak area of Tm -PxB-1 diminished overnight in acidified PBS pH 2.5 by 18%, accompanied by a rise to 4% of the signals for Tm -PxB-2, 1.3% of Tm -PxB-3, and 0.7% of Tm -PxB-4, respectively. Tm -PxB-1 degraded partially to the other three PxBs, most likely by decarboxylation and dehydrogenation in acidic conditions. Therefore, the formation of the pyPxB could be a result of degradation of Tm -PxB-1 during acid extraction processes. To exclude Tm -PxB-2 to be an artefact of purification, freshly harvested senescent nasturtium leaves were investigated under ambient MS conditions using leaf spray ionization ( Fig. 16S , Supporting Information). In Leaf Spray MS, leaf samples are cut prior to analysis and placed directly in front of the ion source, which allows for the direct analysis of the leaf composition and to answer the question, which of the identified PxBs were native ingredients of the plant ³² ,  ³³ . Using this method, all four PxBs were detected, with Tm -PxB-1 being the most abundant. Sodium adducts of Tm -PxB-2 accounted for 92%, Tm -PxB-3 for 9%, and Tm -PxB-4 for 67% of signal intensities, related to the intensity of [M + Na] ⁺ Tm -PxB-1. For Tm -PxB-2, besides an [M + Na] ⁺ ion, also [M + K] ⁺ and [M + 2 K] ⁺ adducts were found; this experiment provides evidence that Tm -PxB-2 is indeed a genuine ingredient of nasturtium leaves.

The in vitro antioxidative potential of the characterized Tm -PxBs was assessed by a FRAP assay ( Fig. 4 ). This assay measures the iron reducing ability of antioxidants, and antioxidant power correlates to the formation of the Fe ²⁺ -(TPTZ) ₂ complex, which can be quantified at 593 nm using a Trolox standard curve. Tm -PxB-1 and Tm -PxB-3 significantly possessed stronger antioxidative activities relative to Trolox, with Tm -PxB-1 displaying two times higher and Tm -PxB-3 three times higher activities than the vitamin E derivative Trolox. Tm -PxB-2 and Tm -PxB-4 exhibited comparable antioxidative strength compared to Trolox. CGA and IQ are known polyphenolic ingredients of nasturtium leaves. Both exhibited stronger effects than Trolox and were in the same range as for the tested PBs and in accordance with the literature (NCCs) ³⁴ ,  ³⁵ . To further investigate the antioxidative effects of PxBs from nasturtium in cellulo , a ROS assay was performed for selected PBs ( Fig. 5 a ). The conversion of an intracellular dye in response to H ₂ O ₂ represents a readout for ROS. Radical scavengers are known for their ability to reduce intracellular ROS production. Demonstrated for Tm -PxB-1 and Tm -PxB-4 in low micromolar concentrations, a reduction of 36% was observed for Tm -PxB-1, while with Tm -PxB-4, a reduction of 39% of ROS production was achieved relative to the positive control (H ₂ O ₂ -treated cells). CGA inhibited ROS production in a comparable manner by 37%. To exclude an influence of the substances on cell viability and therefore also ROS production, cell viability was also assessed. No significant influence on cell viability due to the treatment of the cells with substances was found ( Fig. 5 b ).

Fig. 4.

 PxBs possess strong antioxidative potential in vitro, as assessed by a FRAP assay. Expressed as Trolox equivalents (µmol TE/µmol), Tm -PxB-1 and Tm -PxB-3 showed 2- and 3-fold higher antioxidative power, respectively, and Tm -PxB-2 and Tm -PxB-4 were comparable to Trolox. CGA and IQ, known ingredients of Tropaeolum , also exhibited significantly stronger activity than Trolox. Values represent the mean ± standard deviation of three independent experiments with three replicates each (ns = not significant, * p < 0.05, **** p < 0.0001).

Fig. 5.

 PxBs of T. majus scavenge ROS in HeLa cells ( a ) without influencing cell viability ( b ). a  Cells were stimulated with the compound (10 µM, Trolox 1 mM) for 24 h, then treated with dye (30 min) and H ₂ O ₂ (30 min). Radical scavenging activity was measured as the ability to prevent the conversion of the dye H ₂ DCF to fluorophore DCF. b   Tm -PxB-1, Tm -PxB-2, Trolox, and CGA did not affect cell viability at 10 µM (1 mM for Trolox). Cell viability was tested after 24 h by a crystal violet staining after stimulation of the compounds under the same conditions. For relative quantification, the mean of control triplicates was set to 100% for each repetition. Values represent the mean ± standard deviation of four independent experiments performed in duplicate (ctrl = control, ns = not significant, **** p < 0.0001).

Urinary tract infections are among the indications of T. majus and are mainly attributed to the antibacterial potency of isothiocyanates; therefore, a combination of enriched extracts of nasturtium herb and horseradish roots is commonly used for treatment ³⁶ . Inhibiting the inflammatory cascade by substances of nasturtium via COX-1 and COX-2 can aid in our understanding of the effects elicited by nasturtium in the treatment of cystitis. Investigating the influence of Tm -PxB-1 and Tm -PxB-4 on COX-1 and COX-2 activity, both exhibited inhibitory effects in low micromolar concentrations, similar to CGA. While for both, Tm -PxB-1 and Tm -PxB-4, an inhibition of COX-2 with an IC ₅₀ below 10 µM was observed, Tm -PxB-1 inhibited COX-1 with an IC ₅₀ above 50 µM, and Tm -PxB-4 with an IC ₅₀ of 8 µM. The IC ₅₀ of CGA on COX-2 was 30 µM and 1 µM for IQ. For COX-1, no IC ₅₀ value could be calculated for neither IQ nor CGA. Comparing COX-1 activity dose-response curves of PxBs to CGA and IQ, only Tm -PxB-4 exhibited a pronounced inhibition of COX-1, with an IC ₅₀ of 8 µM. Moreover, IC ₅₀ values on COX-2 inhibition showed a comparable effect of IQ and PxBs, and a presumable stronger inhibitory potential than CGA ( Fig. 6 a ). On the protein level, Tm -PxB-4 significantly diminished COX-2 expression in LPS-stimulated macrophages, as shown earlier for a PxB from Urtica dioica ¹⁵ . In contrast, for CGA, IQ, and Tm -PxB-1, no significant influence on COX-2 expression was found ( Fig. 6 b, c ). Cell viability was measured for all samples and showed that compound treatment did not affect cell viability at the tested concentrations ( Figs. 17S and 18S , Supporting Information). T. majus L. herb has been reported to possess inhibitory effects on the activity of COX-1 ²⁶ . Further investigations on the constitution of the herb extracts responsible for COX-1 inhibition and possible effects on COX-2 have not been examined yet. IQ as an active flavonoid is known for its inhibitory potential against mainly COX-2 ³⁷ . Its ability to downregulate COX-1 activity has only been shown for high concentrations, resulting in IC ₅₀ values above 100 µM ³⁸ . By discovering further potent COX-1 and, in particular, COX-2 inhibitors in nasturtium leaves, PxBs expand the phytochemical repertoire of this medicinal plant.

Fig. 6.

 a  In vitro dose-dependent inhibition of COX-1 and COX-2 activity by Tm -PxB-1, Tm -PxB-4, CGA, and IQ. For IC ₅₀ determination, data of three independent experiments were calculated as percentages of the initial activities and normalized to the positive control with 100% activity. Error bars represent the mean ± SEM (n. d. = SEM was not calculated because of a poor fit of the nonlinear regression). b, c  COX-2 expression of LPS and compound-stimulated J774A.1 macrophages. For determination of COX-2 expression, cells were stimulated with LPS (0.5 µg/mL), and compounds, as indicated, for 4 h before cell lysis and Western blotting. b  Quantitative analysis of COX-2 protein levels was normalized to the whole lane protein and was calculated relative to the LPS-stimulated control. Values represent the mean ± standard error of the mean of three independent experiments (ns = not significant, * p < 0.05, **** p < 0.0001). c  One representative example is shown.

The catabolites accumulating due to the programmed degradation of chlorophyll, the PBs, have long been regarded as irrelevant side products of a detoxification process. Research of the last few years, however, shows a different picture. In particular, yellow chlorophyll catabolites, PxBs, have emerged as interesting bioactive natural products, which are relevant ingredients of medicinal plants. Investigating T. majus , we identified four PxBs, among which one was determined as a pyro-PxB lacking a carboxyl function. We further show that these PxBs can interconvert under acidic conditions. Using leaf spray MS, however, we show that all four identified PxBs occur genuinely in the plant. Moreover, we demonstrate bioactivities related to the use of T. majus in phytomedicine, i.e., antioxidative and anti-inflammatory activities. The identification and investigation of PxBs in T. majus is another important step in establishing an overlooked family of natural products, the PBs, as relevant phytochemicals in medicinal plants.

Material and Methods

General experimental procedures

NMR spectra were recorded on an Avance III HD 500 MHz NMR spectrometer from Bruker BioSpin equipped with a CryoProbe Prodigy broadband probe using d ₆ -DMSO as the solvent. Data were processed using MestreNova 14.1.1.

ESI-MS and MS ² : HR-MS were measured in the LMU Department of Chemistry MS facility; data were processed with Xcalibur 4.1 from Thermo Scientific. MS ² fragmentations and leaf spray mass spectrometry were measured using a Thermo Scientific Q Exactive mass spectrometer and generated with Xcalibur 4.1 and mMass ³⁹ .

Analytical HPLC analysis was performed using either a short or long method for crude extracts with an Agilent 1260 Infinity II LC system with a 1260 Infinity degasser, an 1100 series quaternary pump, an 1100 series diode array detector, and an Agilent Poroshell column 120EC-C18 4 µm 4.6 × 150 mm with a connected Phenomenex ODS 4 × 3 mm i. d. pre-column. The injection volume was 100 µL. The solvent system for long method was mobile phase A = ammonium acetate buffer 10 mM pH 7, B = ACN, flow 0.5 mL/min. Solvent composition was 0 – 5 min 5% B, 5 – 45 min 5% to 60%, 45 – 51 min 60% to 100%, 51 – 53 min 100%, 53 – 55 min 100% to 5%. The short method was used for pure samples or intermediate check up with a solvent system: mobile phase A = ammonium acetate buffer pH 10 mM pH 7, B = ACN, flow 0.5 mL/min, and solvent composition 0 – 2 min 5% B, 2 – 17 min 5% to 100%, 17 – 20 min 100%, 20 – 22 min 100% to 5%.

Semipreparative HPLC was carried out using a Büchi Pure C-830 with a prep HPLC pump 300 bar, fraction collector, prep sample injection valve, and diode array detector. The column was a Luna C18, 5 µm, 250 mm × 21.2 mm, with a Phenomenex pre-column C18 15 × 21.2 mm and the method comprised mobile phase A = phosphate buffer 0.02 M, pH 7, B = MeOH, flow 18 mL/min with solvent composition 0 – 5 min 15% B, 5 – 10 min 15% to 30% B, 10 – 40 min 30% to 55% B, 40 – 42 min 55% to 100% B.

Low-resolution LC-MS was conducted using an Agilent 1100 SL system (G1313A ALS, G1316A COLCOM, G1316A VWD, G1312A Bin Pump) coupled to a Bruker Daltonik HCTultra PTM Discovery system (ESI mode). An Agilent Poroshell column 120EC-C18 4 µm 4.6 × 150 mm with connected Phenomenex ODS 4 × 3 mm i. d. pre-column was used. Solvents were degassed in an ultrasonic bath and samples filtered by a PTFE syringe filter (Ø 0.45 µm). Gradient consisted of A = water + 0.1% FA, B = ACN + 0.1% FA, flow 0.5 mL/min, with a solvent composition of 0 – 2 min 5% B, 2 – 17 min 5% to 100% B.

DMSO stocks of isolated substances were prepared from isolated or purchased substances and stored at − 20 °C. Concentrations of Tm -PxBs were calculated by log ε (426 nm) = 4.51, PleB by log ε (312 nm) = 4.23 using a Thermo Spectronic Genesys 5 (336 001) UV-Visible spectrophotometer, λ _max [nm] ( ε rel) ³ ,  ⁴ . Peak purity was tested using Agilent OpenLab CDS software [UV peak purity (p) in parts per million]: Tm -PxB-1 p = 999.42; Tm -PxB-2 p = 867.07; Tm -PxB-3 p = 996.67; Tm -PxB-4 p = 999.86. Varying peak purity is a result of an intrinsic 8 ² -epimerization of PBs ⁴⁰ , as observed for Tm -PxB-2.

HPLC grade MeOH, EtOH, acetonitrile (MeCN), and HCl were purchased from VWR International GmbH, and ultrapure water (18 MΩ · cm ⁻¹ ) was from a Millipore S. A.S Milli-Q Academic system (18.2 MΩ · cm ⁻¹ ). d ₆ -DMSO was obtained from Eurisotop, hydrogen peroxide (30%) from Bernd Kraft, TPTZ, iron(III) chloride (FeCl ₃ ), and isoquercitrin (purity 98.2%, CP) were from Merck. Chlorogenic acid (purity 98.3%, HPLC) and DMSO (purity ≥ 99.99%, GC) were from Carl Roth. Accutase and LPS were from Sigma-Aldrich. Trolox (purity 100%, HPLC) was from Cayman chemical and 2′,7′-dichlorodihydrofluoroscein diacetate (H ₂ DCF-DA) was from Thermo Fisher. COX Fluorescent Inhibitor Screening Kit was from Cayman Chemicals and SepPak Plus C18 cartridges from Waters Associates. DMEM medium was from PAN-Biotech and fetal calf serum (FCS) was from PAA Laboratories GmbH. COX-2 monoclonal antibody was purchased from Transduction Labaratories and peroxidase conjugated secondary antibody was from abcam.Cell culture. HeLa cells were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) and cultured in DMEM medium supplemented with 10% FCS. J774A.1 cells were a kind gift from Prof. Veit Hornung from the Gene Center Munich and were cultivated in DMEM supplemented with 10% FCS and 1% sodium pyruvate. Both cell lines were cultivated at 37 °C under a 5% CO ₂ atmosphere and split in a 1 : 10 ratio every 3 to 4 days.

Plant material

Senescent fresh leaves of T. majus were collected in the botanical garden of Munich-Nymphenburg (accession number F/0225, determined by Dr. Andreas Gröger, curator of the Botanical Garden Munich-Nymphenburg) and stored at − 20 °C until further use. Spinach ( Spinacia oleracea ) was purchased from a local supermarket.

Extraction and isolation of Tm -PxBs from fresh senescent leaves

PBs were isolated as described earlier according to Karg et al. ¹⁴ with minor adjustments. In brief, 100 g of frozen senescent leaves were ground with 250 mL of extraction mixture consisting of 60% MeOH, 40% PBS 50 mM pH 5.2 using a Braun hand blender Model MR 5000 in a 1000-mL beaker. The mixture was incubated at RT in the dark for 1 h, then filtered through a pleated filter paper with a Ø of 400 mm. The solid residue was washed twice with 200 mL of solvent mixture and filtered again. With the addition of 50% AcOH, the solution was acidified to pH 3.5 and stirred at RT in the dark overnight. Before further purification with semipreparative HPLC, the extract was concentrated to 20 mL, centrifuged (1000  g , 5 min), and filtered. The procedure was repeated for another 100 g of leaves. Fractions containing PBs were combined and repurified by semipreparative HPLC after evaporating and redissolving in 20/80 MeOH/potassium phosphate buffer 100 mM pH 7. The purity of isolated PBs was confirmed by analytical HPLC, then the compounds were dissolved in ACN/potassium phosphate buffer (pH 2.5) 20/80 and stirred overnight. Using SPE (Sep-Pak-C18 cartridge 5 g), pure PBs were eluted with ACN and lyophilized. The stability of the purified compound after acidifying was controlled by HPLC. For further use, DMSO stocks were prepared and stored at − 20 °C.

Chemoprofiling of Tropaeolum majus leaves

A piece of approximately 1 cm ² each of four different yellow leaves of T. majus was cut out ( Fig. 2S , Supporting Information) and ground with 200 µL of MeOH in a mortar, diluted 1 : 11 with potassium phosphate buffer 100 mM pH 7 (20/80), and centrifuged. A 100-µL portion of the extract was analyzed by analytical HPLC. PxBs were assigned by their UV/Vis spectra from a diode array detector, and their relative content was assessed by their HPLC peak areas relative to the sum of the peak areas at 420 nm.

Spectroscopic data

MS/MS fragmentation is presented below; further spectroscopic data is presented in detail in the Supporting Information.

MS/MS Tm -PxB-1 (hcd10): m/z (%) = 663.27 (5, [M + H] ⁺ , C ₃₄ H ₃₉ O ₁₀ N ₄ ⁺ ); 645.25 (75, [M + H-H ₂ O] ⁺ ); 627.24 (32, [M + H-H ₂ O-H ₂ O] ⁺ ); 601.26 (100, [M + H-H ₂ O-CO ₂ ] ⁺ );

MS/MS Tm -PxB-2 (hcd14): m/z (%) = 619.26 (2, [M + H] ⁺ , C ₃₃ H ₃₉ O ₈ N ₄ ⁺ ); 601.27 (100, [M + H-H ₂ O] ⁺ ); 583.26 (4, [M + H-H ₂ O-H ₂ O] ⁺ );

MS/MS Tm -PxB-3 (hcd10): m/z (%) = 647.27 (5, [M + H] ⁺ , C ₃₄ H ₃₉ O ₉ N ₄ ⁺ ); 629.26 (56, [M + H-H ₂ O] ⁺ ); 611.25 (34, [M + H-H ₂ O-H ₂ O] ⁺ ); 585.27 (100, [M + H-H ₂ O-CO ₂ ] ⁺ );

MS/MS Tm -PxB-4 (hcd10): m/z (%) = 629.26 (71, [M + H] ⁺ , C ₃₄ H ₃₇ O ₈ N ₄ ⁺ ); 611.25 (62, [M + H-H ₂ O] ⁺ ); 585.27 (100, [M + H-CO ₂ ] ⁺ ).

Comparison of retention times

A fresh spinach leaf was ground with 200 µL of MeOH in a mortar, diluted 1 : 11 with potassium phosphate buffer 100 mM pH 7 (20/80), and centrifuged. A DMSO stock of isolated PleB was diluted in phosphate buffer 100 mM pH 7 (20/80). Samples were analyzed by analytical HPLC.

Leaf spray

Leaf spray analysis was performed according to a published protocol with minor modifications using a Thermo Scientific Q Exactive ⁴¹ . Yellow senescent leaves were freshly collected and cut into triangles using scissors. The tips of the leaf triangles were mounted in front of the MS inlet. During the measurement, approx. 10 µL of MeOH were dropped onto the leaf triangles several times.

Stability of Tm- PxB-1 in extraction solvents

To test the stability of Tm -PxB-1 during extraction processes, Tm -PxB-1 was dissolved in either PBS of different pH values (pH 7/5.2/2.5) or AcOH (pH 3.5), and peak areas were analyzed by analytical HPLC after 1 h or 24 h of incubation at RT. Relative peak areas were calculated relative to HPLC peak areas of freshly prepared Tm -PxB-1 in PBS pH 7, which was immediately injected. PxBs were identified by their UV spectrum and retention times. Tm -PxB-1 samples in PBS pH 2.5 overnight were additionally analyzed by LC-MS.

Biological assays

Ferric reducing antioxidant potential assay

The FRAP assay was performed as described by Karg et al. ¹⁴ with minor modifications. Briefly, freshly prepared FRAP reagent was added to 100 µM of compounds and different concentrations of Trolox for 5 min at 37 °C in a 96-well plate. The FRAP reagent consisted of 10 vol 300 mM acetate buffer pH 3.6, 1 vol 10 mM TPTZ in 40 mM HCl and 1 vol 20 mM iron(III) chloride. Antioxidants reduced the Fe ³⁺ -(TPTZ) ₂ complex to intense blue Fe ²⁺ -(TPTZ) ₂ complex with an absorption maximum at 593 nm. A calibration curve was established by measuring the absorbance of different Trolox concentrations. The antioxidant power of the compounds was calculated relative to Trolox and expressed as Trolox equivalents.

In cellulo reactive oxygen species assay

The intracellular ROS assay was conducted according to Karg et al. ¹⁴ . In short, 1 × 10 ⁵ HeLa cells/mL in 100 µL were seeded in 96-well plates and preincubated for 24 h. After treatment with Tm -PxB-1 (10 µM), Tm -PxB-4 (10 µM), CGA (10 µM), Trolox (1 mM), or DMSO control for 24 h, the medium was removed, H ₂ DCF-DA (10 µM) was added, and the plate was incubated for 30 min. After washing with PBS, cells were incubated with hydrogen peroxide (1 mM) for 30 min, and the formation of a highly fluorescent 2′,7′-dichlorofluorescein (DCF) by intercellular oxidation of H ₂ DCF-DA was measured at the end of the incubation time by a Tecan SpectraFluor plus microplate reader (excitation wavelength 485 nm; emission wavelength 530 nm). Data were normalized to the hydrogen peroxide-treated DMSO control and cell viability was assayed by a crystal violet staining, measured with a spectrophotometer at 590 nm. The number of viable cells was normalized to a DMSO control.

Cyclooxygenase inhibition assay

The COX inhibition assay was performed using the Cayman Chemicals COX Fluorescent Inhibitor Screening Kit according to the manufacturerʼs instructions. In this assay, arachidonic acid is converted to hydroperoxyl endoperoxide (PGG2) by a COX component of the bifunctional enzymes COX-1 and COX-2. Followed by the reduction of PGG2 to prostaglandin H2 by the peroxidase component and 10-acetyl-3,7-dihydroxyphenoxazine (ADHP), highly fluorescent resorufin is generated. Its fluorescence can be measured with an extinction wavelength of 535 nm and an emission wavelength of 595 nm. In brief, ovine COX-1 or human recombinant COX-2 enzymes were incubated with Tm -PxB-1, Tm -PxB-4, CGA, or IQ in different concentrations, or buffer as a positive control with 100% initial COX activity, for 5 min. Wells without enzymes served as the negative control. The fluorescence intensity was measured 2 min after adding arachidonic acid. After subtracting the negative control, percentages of initial activities were normalized towards the positive control with 100% initial activity. IC ₅₀ values were calculated by nonlinear regression analysis utilizing GraphPad Prism 9.

Cell treatment for Western blot and cytotoxicity analysis

J77A.1 cells were detached with accutase for an incubation time of 20 min. Then, 2 × 10 ⁵ HeLa cells/well in 1000 µL were seeded in 12-well plates and were allowed to attach for 24 h. Cells were stimulated with LPS (0.5 µg/mL) and compounds as indicated for 4 h.

Western blot analysis

For cell lysis, cells were washed twice with ice-cold PBS. Then, 100 µL of lysis buffer (50 mM Tris/HCl, 150 mM NaCl, 1% Nonidet NP-40, 0.25% sodium deoxycholate, 0.1% SDS) were added and lysates were frozen at − 80 °C until further use. Lysates were thawed on ice and centrifuged (14 000 rpm, 10 min, 4 °C) to remove cell debris. Protein concentrations were quantified by the Bradford assay and protein concentrations were quantified by linear regression analysis with BSA dilutions as protein standards. To adjust protein concentrations, each sample was diluted with SDS. Proteins were denatured at 95 °C for 5 min and were separated by SDS-PAGE in electrophoresis buffer (100 V, 21 min then 200 V, 43 min) on discontinuous polyacrylamide gels. Gels were combined by a gradient separation gel (4 – 20%) and a stacking gel. By tank blotting (100 V, 90 min, 4 °C), proteins were transferred to polyvinylidene difluoride (PVDF) membranes that were equilibrated before in tank buffer. Afterwards, membranes were blocked with 5% nonfat dry milk powder in PBS for 2 h. Primary antibody was added overnight at 4 °C (1 : 1000). Membranes were washed four times with TBS-T for 5 min each, before the secondary HRP-coupled antibody was added for 2 h (1 : 1000). A four-time washing procedure was repeated prior to the incubation with ECL solution and chemiluminescence detection with a ChemiDoc Touch Imaging System. By comparison with Page Ruler Plus Prestained Protein Ladder, bands were assigned. Protein levels were analyzed by Image Lab Software and were normalized to the total protein level.

Cytotoxicity determination

In parallel, cell viability under the conditions of the Western plot experiment was assessed by two methods: (1) treated cells were washed with PBS, detached with 200 µL accutase for 20 min, then diluted with 500 µL culture medium; (2) 350 µL of the cell suspension were transferred into a ViCell XR cell counter and cell viability was measured by trypan blue staining. Additionally, the same amount of cell suspension was used for flow cytometry. Therefore, cells were centrifuged (1000 rpm, 5 min, 4 °C) in precooled FACS tubes, washed twice with ice-cold PBS, and resuspended in 200 µL PBS. Directly after, cells were kept in the dark on ice while flow cytometric analysis was conducted on a BD FACS Canto II (BD Biosciences). To determine viable cells in the samples, cells were gated for the living cells in the DMSO control by forward scatter (FCS) and side scatter (SSC) using FlowJo 7.6 software (BD Biosciences).

Statistical analysis

Results display the mean of at least three independent experiments (mean ± standard deviation), each replicated in at least three repeats, if not stated otherwise. Statistical significance was calculated by one-way analysis of variance with post hoc analysis using Dunnettʼs multiple comparison test. All statistical analyses were performed with GraphPad Prism 9.

Contributorsʼ Statement

Conception and design of the work: SM, AV, TM; data collection PF, CN; analysis and interpretation of the data: PF, CN, AV, TM, SM; statistical analysis: PF, CN; drafting the manuscript: PF, CN, AV, TM, SM; critical revision of the manuscript: PF, CN, AV, TM, SM.

Funding Statement

S. M. gratefully acknowledges funding from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation; Project-ID 448289381).