Phylloxanthobilins are Abundant Linear Tetrapyrroles from Chlorophyll Breakdown with Activities Against Cancer Cells

: Linear tetrapyrroles, called phyllobilins, are obtained as major catabolites upon chlorophyll degradation. Primarily, colorless phylloleucobilins featuring four deconjugated pyrrole units were identified. Their yellow counterparts, phylloxanthobilins, were discovered more recently. Although the two catabolites differ only by one double bond, physicochemical properties are very distinct. Moreover, the presence of the double bond seems to enhance physiologically relevant bioactivities: in contrast to phylloleucobilin, we identified a potent anti-prolifer-ative activity for a phylloxanthobilin, and show that this natural Introduction statistical analyses were processed with GraphPad Prism 7.05. The authors would like to thank Kerstin Schmid and Claudia Glas for experimental assistance, and Prof. Susanne S. Renner for her support with preparing the voucher specimen for the herbarium. P. Wang is supported by a Fellowship from the Chi-nese Scholarship Council. S. Moser acknowledges financial support from the CUP Mentoring program. Open access funding enabled and organized by Projekt DEAL.


Introduction
Despite its visibility, the biochemical degradation of the green plant pigment chlorophyll (Chl) has remained unresolved for a long time, until abundant linear tetrapyrroles, now called phyllobilins (PBs), were discovered as stable degradation products accumulating in the vacuoles of the plant cell. [1] In the meantime, Chl breakdown has revealed many of its mysteries, and we now have a well-defined picture of the biochemical program of Chl breakdown, called pheophorbide a oxygenase (PaO)/phyllobilin pathway. The most commonly identified "final" breakdown product is a 3 2 -hydroxylated phylloleucobilin (PleB), 3 2 -OH-PleB. Colorless PleBs were the first PBs to be discovered and were believed to represent the "last stage" of Chl breakdown.
product induces apoptotic cell death and a cell cycle arrest in cancer cells. Interestingly, upon modifying inactive phylloleucobilin by esterification, an anti-proliferative activity can be observed that increases with the chain lengths of the alkyl esters. We provide first evidence for anti-cancer activity of phyllobilins, report a novel plant source for a phylloxanthobilin, and by using paper spray MS, show that these bioactive yellow chlorophyll catabolites are more prevalent in Nature than previously assumed.
The formation of PxBs in plants has not yet been fully clarified, but it is assumed that plants with PxB content show an endogenous "oxidative activity", likely caused by enzymes. [4] In contrast to PleBs, PxBs feature a double bond system extending from ring C to ring D, resulting not only in an intense yellow color, but also in interesting chemical properties. PxBs were shown to be reversible photoswitches and to dimerize depending on the lipophilicity of the environment in a formal [2+2] cycloaddition. [5] Chl breakdown has been and still is regarded primarily as a detoxification process; the discovery of the PxBs, however, points towards yet to be elucidated physiological roles of these metabolites for the plant. Up to the stage of the PleBs, the efforts of the PaO/phyllobilin pathway, using metabolic energy and specific enzymes to detoxify photoactive Chl, yields nonphotoactive, more water-soluble molecules with four isolated tetrapyrrole units. The discovery of PxBs, however, contradicts the detoxification paradigm. PxBs were found to be photoactive compounds [5] and are, compared to PleBs, less hydrophilic, seemingly reverting the efforts of the detoxification pathway and raising the question of a potential relevance of PxBs in physiological roles for the plant.
For humans, PxBs are suspected to have yet un-identified bioactivities, too. [6] We could recently show that the PxBs from Echinacea purpurea have a very potent in vitro antioxidative activity, a high radical scavenging potential in human cells, and can protect the cells from oxidative stress. Furthermore, metabolic studies indicated the PxBs to be stable in the cell. [7] An-Since PxBs appear to be overlooked bioactive constituents of plants, and compounds with structures related to PBs, such as bilirubin, have a reported anti-cancer activity, [9] we set out to test the cytotoxicity of PBs against cancer cells. We focused on the most common PB motif, the 3 2 -hydroxylated core structure, 3 2 -OH-PleB (1 and 3), and its PxB oxidation product 2. The PxB (2) showed promising cytotoxicity and apoptosis-inducing properties in bladder and breast cancer cells. The PleB precursors 1 and 3, in contrast, were revealed not to be toxic against cancer cells; we could show, however, that esterification of the propionic acid side chain by chemical synthesis led to tunable cytotoxicity, which could be partially explained by uptake studies.

Isolation and Structure Elucidation of PleB (3) and PxB (2) from Plane Tree (Platanus occidentalis)
The analytical HPLC analysis of a senescent plane tree leaf revealed two signals with UV spectra characteristics for phyllobilins ( Figure 1, Figure S1). One signal was tentatively identified as PleB with absorption maxima at around 240 nm and 312 nm and one as a PxB featuring an additional maximum at 426 nm. [3] The Po-PleB (3) was isolated from 300 g of leaves as described in the experimental section, giving 50 mg of pure 3. In order to obtain large amounts of 2, the latter was synthesized from 3 via light assisted oxidation on silica gel. [10] Using the published protocol, 2 mg of the PxB (2) were obtained from 10 mg of 3. HRMS analysis of purified compounds allowed for the deduction of the molecular formulae as C 35 H 41 O 8 N 4 for 3, and C 35 H 39 O 8 N 4 for 2, revealing them to have an identical molecular composition as PleB (1) and PxB (2) from katsura tree leaves. PBs from katsura tree, Cercidiphyllum japonicum, have been thoroughly characterized by spectroscopic methods; the major PleB (1) was identified as 3 2 -OH PleB. [2] Furthermore, the Z and E conformers of 3 2 -OH-PxB (2) were identified. [3]  In a co-elution experiment, in which isolated pure Po-PleB (3) and Cj-PleB (1), and a mixture of the two pure compounds, were applied to analytical HPLC, the two PleB signals showed similar but not identical retention times ( Figure S2 A-C). Since the mass spectra of the two compounds revealed them to have an identical molecular formula, we conclude that the two PleBs 1 and 3 are epimers featuring a different configuration at the Eur. J. Org. Chem. 2020, 4499-4509 www.eurjoc.org 4501 stereocenter at C16. Two "classes" of PleBs, i.e. C16 epimers, result from the stereospecificity of an enzyme that introduces a new stereocenter in the PleB precursor primary phyllolumibilin (pPluB) either in the "n" or "epi" configuration, depending on the plant species. [11] The pPluB is then hydroxylated resulting in a 3 2 -OH PluB, which is converted into the respective PleB by acid-induced tautomerization in the vacuoles of the senescent plant cell. [12] The elemental composition of Po-PxB (2) was deduced via HRMS as C 35 H 39 O 8 N 4 and revealed the structure to be identical with Cj-PxB (2) from katsura. A co-elution experiment was performed, in which the isolated compounds Cj-PxB (2), Po-PxB (2), and a mixture of both compounds, were analyzed by analytical HPL-chromatography. The chromatograms showed identical peaks and chromatographic characteristics ( Figure S2 D-F). Moreover, these findings were confirmed by 1 H NMR spectroscopy. By comparing the 1 H NMR spectrum of Po-PxB (2) with the spectrum of Cj-PxB (2), [3] the same typical signals for a tetrapyrrolic PxB stood out: the characteristic signal pattern for the spin system of the peripheral vinyl group at the intermediate field; four singlets at high field of the methyl groups H 3 C2 1 , H 3 C7 1 , H 3 C13 1 , and H 3 C17 1 near 2 ppm, and one singlet of the methyl ester group at 3.7 ppm. Moreover, one singlet at the low field for the formyl group was detected, as well as the characteristic signal for the additional double bond in a PxB core in Z conformation, [3] a singlet of HC15 at 6.19 ppm ( Figure S5, Table S2). In conclusion, the spectroscopic data clearly show the PxB from Platanus to be identical to the PxB (2) from Cercidiphyllum japonicum.

Analysis of PBs in Senescent Leaves by Paper Spray Mass Spectrometry
Since the discovery of the first linear chlorophyll catabolite, a PleB, 29 years ago in the leaves of barley, [1] the list of PleBs identified from different plant species has grown considerably; PleBs were shown to occur in a large structural variety arising from the conjugation of the tetrapyrrole core to hydrophilic residues.
With the first PxB (2), detected in leaves of katsura, [3] the spectrum of chlorophyll catabolites was extended beyond PleBs. Following the PxB (2) from katsura, investigations on other plant species have yielded PxBs; PxBs have been identified in leaf extracts of Egeria densa, [13] Tilia cordata, [14] plum tree, [15] and wych elm. [16] The abundances of PxBs compared to their PleB precursors, however, appeared to be lower. Exceptions to this observation are the PxBs discovered in the medicinal plant Echinacea purpurea, in which PxBs were detected in unprecedented abundance and diversity. [7] These findings raised the question of a broader occurrence of PxBs; since many plant species have been investigated before the discovery of PxBs, only PleBs are reported for those. The most common 3 2 -OH PleB has been identified, among others, in Cercidiphyllum japonicum, spinach (Spinacia oleracea), and Spathiphyllum wallisii. [17] Taken into account that PxBs hold a large promise of yet unexplored bioactivities, we re-investigated Spinacia oleracea and Spathiphyllum wallisii, reportedly containing 3 2 -OH-PleB (1 and 3, respectively), but specifically looking for the presence of the 3 2 -OH PxB (2).
A characteristic feature of PleBs, which originally led to the term "rusty pigments", is that when applied to thin-layer chromatography on silica gel, the colorless PleB spots developed first into yellow spots, and eventually into rust-colored spots under exposure to daylight. [3] Since the oxidation of PleBs to PxBs is influenced by air and light, higher apparent abundances of PxBs might be artifacts of the extraction procedure. We, therefore, used a combination of paper spray and leaf spray mass spectrometry which allowed the rapid and direct analysis of the leaf tissue without any sample preparation. Paper spray as well as leaf spray are rather new applications in the field of ambient mass spectrometry, which were first described by Liu et al. in 2010 [18] and 2011. [19] In general, mass spectrometry has been developed into an efficient tool for identification and structural elucidation of phyllobilins with known modification patterns. The core structure of the linear tetrapyrroles has been characterized extensively for more than 20 different examples leading to the creation of an MS database for phyllobilins. [20] Due to characteristic fragmentation patterns of phyllobilins, MS and MS/MS are straightforward tools for the characterization of this family of linear tetrapyrroles. [21] Paper spray MS analysis of freshly collected, senescent leaves ( Figure 2 and Figure 3) clearly revealed the presence of both types of chlorophyll catabolites, colorless PleB and yellow PxB, in the intact senescent leaves of Cercidiphyllum japonicum as well as Platanus occidentalis, Spinacia oleracea, and Spathiphyllum wallisii. PleBs and PxBs were identified with respect to the fragmentation behavior of their isolated molecular ions in the gas phase. [21] Collision induced dissociation (CID) of the monoisotopic, potassiated molecular ions ([M + K] + ) of 3 2 -OH PleB (1 and 3) at m/z = 683 showed diagnostic fragments at m/z = 651, which correspond to the characteristic loss of methanol (-32 Da). Interestingly, the two most abundant fragment ions of monoisotopic, potassiated molecular ions ([M + K] + ) of 3 2 -OH PxB (2) at m/z = 681 correspond to the loss of methanol (-32 Da) as well as to the loss of water (-18 Da). The paper spray confirms the occurrence of PxB (2) in all four investigated plant species, for which only the PleB precursor (1 or 3) was reported before. Figure 2. Example of a paper spray setup. The yellow, freshly collected senescent leaf from katsura tree (Cercidiphyllum japonicum, left) was cut using razor blade (center). The 0.5 cm × 3 cm slice was wrapped using a tapered piece of filter paper and mounted in front of the MS inlet (right). A potential of 3.5 kV was applied to the wrapped leaf, and the MS n data were recorded within 30 s while applying 20 μL of methanol.

Cytotoxic Potential of Phyllobilins
Structurally, phyllobilins bear a remarkable resemblance to heme-derived bilins; [22] whereby PxB, at least structurally, appears to be the bilirubin plant counterpart. [8] The bilins as well as the phyllobilins were long thought to be mere waste products, generated in a detoxification process of heme; [23] this assumption still holds true for the products of the chlorophyll degradation pathway. For the bilins, this paradigm has been disproved; e.g. bilirubin was shown to act as an important physiological antioxidant and to possess cytoprotective potential against various diseases. [24] In addition, dependent on the concentration as well as the state of redox homeostasis in the cell, bilirubin can also act as a cytotoxic agent. [25] This cytotoxicity plays a role in the anti-cancer effects attributed to the heme catabolite, among which anti-tumoral effects on human adenocarcinoma, [26] and apoptosis-inducing activity in colon cancer cells were described. [9] The growing evidence for physiological activities of the bile pigments fueled the search for relevant properties of phyllobilins. In a first report, Müller et al. were able to demonstrate in vitro antioxidative activities for PleBs isolated from the peels of apples and pears. [27] For the PxBs, we recently reported their potent antioxidative activity in vitro; in addition, PxBs turned out to be rapidly taken up by cells, to be potent ROS scavengers in cells and to possess the ability to protect cells from oxidative stress. [7] Further bio-relevant effects on cells, however, still remain to be elucidated. In this study, we set out to probe the cytotoxic potential of phyllobilins on cancer cells; in a first experiment, we tested the PleB from katsura tree Cj-PleB (1) and its epimer from plane tree leaves Po-PleB (3) on two different human cancer cell lines, the highly invasive epithelial breast cancer MDA-MB-231 cell line and the urinary bladder carcinoma T24 cell line. 1 and 3 showed no significant inhibition of cell proliferation over 72 h in the tested concentration range of 1 to 100 μM ( Figure 4A). For testing the cytotoxic potential of PxB (2), Cercidiphyllum japonicum was used as a source for the isolation of 1 and its subsequent oxidation; hence, only Cj-PxB (2) was used as test reagent and further defined as PxB (2), since Cj-PxB (2) and Po-PxB (2) are identical compounds due to the unsaturation of the stereocenter at C16 of the PleB (1 or 3). In contrast to the PleB precursors (1 or 3), 2 exhibited a potent anti-proliferative effect at low micromolar doses in both cell lines with calculated IC 50 values of 7.0 μM against MDA-MB-231 cells and 4.6 μM against T24 cells ( Figure 4B).

Phylloxanthobilin Inhibits Colony Formation
Having established that 2 inhibits the proliferation of two different cancer cell lines, we further investigated this effect by a colony formation assay. This assay assesses the ability of single cells to grow and form colonies after a short time exposure to the compound by a crystal violet staining. The results showed a steep decline of colony formation in comparison to the DMSO vehicle control after brief exposure to 10 μM 2, indicating 2 to inhibit also long-term survival and proliferation of cancer cells ( Figure 4C).

Phylloxanthobilin Induces Apoptosis and a G2/M Cell Cycle Arrest
Furthermore, we assessed the cytotoxic potential of PxB (2) by flow cytometry, using a commercial Annexin V/PI staining kit, which allows for the distinction between different types of cell death. FITC conjugated Annexin V targets apoptotic cells and counterstaining with PI identifies cells undergoing early apoptosis (A+/PI-) or late apoptosis (A+/PI+). In contrast, cells with PI positive and Annexin V negative signal are related to a nonapoptotic, necrotic cell death. The experiment showed that PxB (2)    In contrast, no effect on the population in the S phase was observed during treatment with 2. Moreover, a significant increase of cells in the subG1 phase in both cell lines was observed, confirming the results of the Annexin V/PI assay that 2 promotes apoptotic cell death ( Figure S7).

Cytotoxicity of PleB Can Be Tuned by Esterification
So far, PleBs have been identified and characterized as approximately 20 different structures, whereby identical PleBs have been isolated from different plant species. All PleBs share the same core structure with four deconjugated pyrrole units. Differences arise from modifications of the rings at four different positions (with the exception of a catabolite from Arabidopsis thaliana, which is speculated to result from an incomplete reduction from Chl b to Chl a [29] ).
Of particular interest is the esterification of the propionic acid side chain at C12; Oberhuber et al. showed that introducing a methyl ester at the PluB stage by chemical synthesis changed the kinetics of the conversion to PleBs, resulting in increased stability of the esterified PluB. [12b] An esterification of the propionic acid side chain of PBs was also discovered to occur naturally, causing PluBs in banana peels to accumulate and, as a consequence, the peel of ripened bananas to luminesce blue. [30] On the PleB level, esterified compounds were identified in leaves of Vitis vinifera [31] and wych elm, [16] and very recently on the PxB level in Epipremnum aureum. [32]  In comparison to the PxB, the PleBs as more polar molecules showed no effect on the proliferation of cancer cells. We therefore modified the structure of the PleB by esterification of the propionic acid side chain aiming to decrease its polarity. For our experiments, we used Cj-PleB (1) as representative compound. We prepared four different esters of the propionic acid side chain with increasing length of the alkyl side chain as described in the experimental section to gradually decrease polarity ( Figure 6). Chemical structures were confirmed by HRMS, ESI MS/MS and NMR measurements ( Figure S8-S12, Table S3,4).
Indeed, the inhibitory effects on the proliferation of MDA-MB-231 and T24 cells were influenced by esterification, with activities increasing with the chain lengths of the alkyl esters ( Figure 7). In MDA-MB-231 cells, the IC 50 value of the methylcompared to the ethyl ester showed a decrease with polarity (46.1 μM for the methyl ester (4) and 30.9 μM for the ethyl ester (5)), and even lower values were observed for the butyl ester (6), as well as for the octyl ester (7); butyl ester (6) exhibited a IC 50 value of 23.6 μM and octyl ester (7)  Thus, we reasoned that polarity might be a factor that plays a role in the observed differences in cytotoxic potential between PleBs.

Uptake of PxB (2), PleBs (1, 3), and Esterified PleBs (4-7) by Cancer Cells
A crucial factor for the efficacy of a therapeutic agent is the delivery into the cell, as most of the substances target intracellular constituents. The ability for small drugs to overcome the natural barrier of the cell, the plasma membrane, is mainly determined by the size and lipophilicity of the compound. Whereas lipophilic small molecules can easily enter the cell e.g. through diffusion or active transport, a weak membrane permeability limits the cellular uptake of hydrophilic or larger molecules. Therefore, a common method to improve bioavailability is to alter the lipophilicity of molecules by generating prodrugs. [33] Although the mechanism of the cellular uptake for phyllobilins remains to be elucidated, the polarity of the different tetrapyrrolic compounds seems to be a factor influencing the cytotoxicity. We, therefore, investigated the uptake of different PBs by cells, using HPLC to quantify the PB signal from cell lysates. This experiment also aimed at clarifying whether the PleB-esters (4-7) act like prodrugs, serving as vehicles of the compounds into the cells. In this case, the esters are cleaved by cellular esterases afterward, leading to an increase of the effective concentration of PleB (1) in the cell compared to the use of the un-esterified compound; it could also be the case, however, that the esterified PleB itself has a cytotoxic activity.
All tested compounds could be found in the lysates of T24 cells, confirming that PBs, in general, can be taken up by cancer cells and are stable under the indicated conditions ( Figure S13). In comparison to the PxB (2), the Cj-PleB (1) and Po-PleB (3) were only detected in minute amounts (Figure 8). This indicates the lack of cytotoxicity of the PleB (1,3) to correlate with higher polarity and lower intracellular compound concentration.
In contrast, the esterified PleBs (4-7) were detected in high concentrations. As expected, the intracellular concentration increased in parallel with apolarity of the compounds, being lowest for the methylated PleB (4) and highest for the butylated PleB (6). For the most apolar octylated PleB (7), however, a decreased intracellular concentration was observed, which might be due to low solubility in medium ( Figure 8).
Interestingly, only minute amounts of hydrolyzed "free" PleB (1) could be observed in cell lysates when treating cells with the PleB-esters (4-7) ( Figure S8). These results indicate the anti- proliferative effects of 4-7 not to depend on the liberation of the "free" PleB (1) in the cell. More likely, the esters appear to exhibit cytotoxic activities themselves, which cannot be explained by a prodrug mechanism. The observed reactivity of the PxB (2), however, surpasses the activity of 4 and 5. Although the IC 50 values of 6, 7, and 2 in T24 cells are in a similar low micromolar range, 6 and 7 showed a significantly lower polarity than the PxB 2 ( Figure 6) and a higher uptake (Figure 8). Nevertheless, neither polarity nor uptake can account for the difference in potency between PleBs and PxB (2). This indicates the structural difference between PleB and PxB, manifested by a double bond, to play a crucial role in inhibiting proliferation of cancer cells.
Whether a PleB-ester acts as a prodrug and the ester gets cleaved by cellular esterases, or the esterified compound itself is the cytotoxic agent, needs yet to be resolved in more detail; our results indicate, however, that only minute amounts of hydrolyzed PleB (1) are present in the cells, even 5 h after treatment.

Conclusion
In this study, we identified plane tree as novel source for 3 2 -OH-PxB (2). This yellow chlorophyll catabolite was shown to be more abundant in Nature than previously thought, as was shown by paper spray MS; we demonstrated the occurrence of PxB (2) in leaves of Spinacia oleracea and Spathiphyllum wallisii, for which only the common 3 2 -OH PleB (1 or 3) has been reported previously. After all, the discovery of PleBs precedes the discovery of PxBs by 17 years, therefore the studies during that time range did not screen for PxBs. Whether PxBs fulfill biological roles for plants, has yet to be investigated.
Here, we could introduce PxB (2) as potent anti-cancer agent in human cells, inhibiting proliferation in the low micromolar range, inducing apoptosis, and causing a G2/M cell cycle arrest. We provide first evidence for anti-cancer activities of PxBs and add another proof that chlorophyll catabolites are more than mere waste products of a degradation pathway.
The discovery of a potent activity for the PxB (2), the fact that the cytotoxicity of PleBs (1,3) on cancer cells is tunable by a simple modification of the compounds, and the abundance of this family of natural products in the plant kingdom, open the door for further investigations of the anti-tumoral potential of the phyllobilins. Future studies will deal with the mechanism behind the potent killing and apoptosis induction in cancer cells and aim at identifying new sources of PxBs, which hold a large promise to reveal yet to be explored bioactivities.
Human bladder cancer cell line T24 and highly invasive human triple negative breast adenocarcinoma cell line MDA-MB-231 were obtained from the Deutsche Sammlung von Mikroorgansimen und Zellkulturen (DSMZ; Braunschweig, Germany) and maintained in DMEM medium supplemented with 10 % fetal calf serum (FCS) and 1 % penicillin/streptomycin. Cells were cultured at 37°C under 5 % CO 2 atmosphere with constant humidity in 75 cm 3 tissue culture flasks.
Plant Material: Katsura trees (Cercidiphyllum japonicum) were located outside of the botanical garden Munich using the tree-finder app https://www.botmuctrees.de/. Senescent leaves were collected in the Maria-Ward-Straße, Munich (48°09′41.0"N 11°29′58.8"E). Senescent leaves of plane tree (Platanus occidentalis) were collected from trees in the Feodor-Lynen Straße at the campus Großhadern of the University of Munich (48°06′48.6"N 11°27′56.2"E). Spinach was obtained from a local supermarket and stored in the dark for 3 days to induce Chl breakdown. Spathiphyllum wallisii was bought at a local garden center. A senescent leaf was collected directly from the plant. Spectroscopy: UV/Vis: Thermo Spectronic Genesys 5 (336001) UV Visible spectrophotometer; λ max in nm (rel. ε). Concentrations of PleB and PleB-esters were calculated using log ε (312 nm) = 4.23, [4] concentrations of PxB were calculated using log ε (426 nm) = 4.51. [3] Paper spray mass spectrometry: Thermo Scientific Orbitrap XL mass spectrometer. The freshly collected senescent (yellow) leaves were cut into 0.5 cm × 3 cm slices using a razor blade. A freshly cut slice was wrapped into a tapered piece of filter paper and mounted in front of the MS inlet at a distance of 0.5 cm. A high voltage potential of 3.5 kV was applied to the wrapped leaf and MS n data were recorded within 30 s while 20 μL of methanol were dropped onto the filter paper.
HRMS were measured at the MS facility of the Department of Chemistry, University of Munich. Data were processed with Xcalibur.
NMR spectra were recorded on an Avance III HD 500 MHz NMR spectrometer from Bruker BioSpin equipped with a CryoProbe™ Prodigy broadband probe using CD 3 OD as solvent. Assignments of 13 C signals determined from 1 H, 13 C heteronuclear correlations in HMQC and HMBC spectra. NMR data were analyzed with Mestre Nova 14.1.1.
During all handling steps, the material was protected from light and temperatures above 37°C were avoided unless stated otherwise.
HPLC Analysis of Plane Tree (Platanus occidentalis) PBs: 16 cm 2 of a senescent leaf of plane tree was mixed with sea sand and 2 mL of MeOH in a mortar. The slurry was centrifuged (1000 rpm, 5 min, 4°C), and a 40 μL aliquot of the supernatant was diluted with 160 μL of phosphate buffer (pH 7.4); a 100 μL aliquot was applied to analytical HPLC. In a co-elution experiment, isolated pure samples Po-PleB (3) and Po-PxB (2), respectively, were applied to analytical HPLC. Cj-PleB (1) and Cj-PxB (2) were isolated from katsura leaves as described below and were analyzed the same way. In addition, a 1:1 mixture of 1 and 3, and of 2 from the two different plant sources, respectively, were analyzed by HPLC.

Large Scale Isolation of PleB from Senescent Leaves of Katsura (Cercidiphyllum japonicum) and Plane Tree (Platanus occidentalis):
Senescent leaves of katsura and plane tree were extracted as follows (optimized protocol for plane tree described): 300 g of frozen leaves were ground in a 5 L stainless steel beaker using a Braun hand blender Model MR 5000 and extracted with hot water (1200 mL). The mixture was filtered through a cotton cloth and the residue was again washed with hot water (500 mL (Table S1). and incubated with FITC conjugated AnnV in binding buffer for 10 min. After washing with binding buffer, PI solution was added and cells were immediately analyzed by a BD FACS Canto™ II cytometer. PI and AnnV negative cells were identified as living cells, PI negative and AnnV positive as early apoptotic cells, PI positive and AnnV positive as late apoptotic, and PI positive and AnnV negative as necrotic cells. Data were processed with FlowJo 7.6 software.

Po-PxB
Cell Cycle Analysis Assay: The cell cycle distribution of T24 and MDA-MB-231 cells was determined by propidium iodide staining and flow cytometry according to the protocol of Nicoletti et al. [28] Briefly, 3 × 10 4 cells were seeded and allowed to attach for one day. Next, cells were treated with 2 (5, 10, 20 μM), and DMSO as a vehicle control and incubated for 24 or 48 h. Subsequently, cells were detached, washed with PBS, and stained with fluochrome solution (50 μg/mL PI in a solution of 0.1 % sodium citrate (w/v) and 0.1 % Triton X-100 (v/v) in deionized water) for 30 min at 4°C in the dark. Cell cycle was analyzed using a BD FACS Canto™ II and data were evaluated using FlowJo 7.6 software.
Cell Uptake Assay: The uptake of PBs was determined by analytical HPLC analysis. 1.5 × 10 6 T24 cells were seeded in 60 mm dishes and incubated for one day. Next, cells were treated with Cj-PleB (1), Po-PleB (3), PxB (2), and PleB-esters (4-7) (80 μM) for 5 h. After washing twice with ice-cold PBS, cells were scraped off in 500 μL of PBS and cell suspension was centrifuged at 13000 rpm for 5 min at 4°C. The supernatant was discarded and the cell pellet was again washed with PBS. Cells were lysed by bead beating and proteins were precipitated with 120 μL of ACN/PBS (20:80) for 1 h on ice. Lysates were centrifuged again and a 50 μL aliquot of the supernatant was analyzed by HPLC. The peak areas of the different compounds were normalized to the peak area of the butylated ester, which was set to 100 %.
Statistical Analysis: Results represent the mean of at least three independent experiments (means ± standard deviation) performed in at least three replicates unless stated otherwise. Statistical significance was carried out by two-way analysis of variance with post hoc analysis using Dunnett's multiple comparison test; all statistical analyses were processed with GraphPad Prism 7.05.