The TLC separations were performed on Merck^® plates, silica gel 60, with fluorescence indicator PF₂₅₄ and an aluminum support of thickness 0.2 mm. The prep-TLC was carried out on 20 × 20 cm size glass plates, 1.0 mm thickness of Merck silica gel 60 and fluorescence indicator PF₂₅₄. The plates were visualized under 254 and 365 nm ultraviolet light or nebulized with sulfuric vanillin solution followed by heating.

For circular chromatography, round glass plates coated with a 2.0 mm thick layer of silica gel 60, PF₂₅₄ containing gypsum, were used. The compounds were separated using hexane-EtOAc as the eluant, and were latter visualized under ultraviolet light (254 and 365 nm).

Purification of compounds by column chromatography was performed using VLC (vacuum liquid chromatography), with columns of appropriate length and diameter necessary for the masses of the samples (Pelletier et al., 1986). Silica gel 60 HF₂₅₄ (70–230 mesh ASTM) from Merck^® and C-18 reverse phase silica were used as the stationary phase. The ratio of silica used to pack the column was approximately 20 times the mass of the sample to be purified.

The specimens of P. pellucida were maintained in the greenhouse facilities of the Institute of Chemistry at the University of São Paulo (IQUSP) under controlled conditions (25 ± 2°C; photoperiod of 12 h; 85 W fluorescent lamps). The Piper and Peperomia species growing in the garden at IQUSP naturally host several arthropods, such as Edessa meditabunda (Hemiptera), Monoplatus sp. (under identification; Coleoptera; Chrysomelidae), and Geometridae sp. (under identification, Lepidoptera) that were collected in the garden and used in biotic plant stress experiments (Supplementary Figure 1). The experiments with herbivores wounding were carried out in cages using five replicates containing three plants with three herbivores each, excepting for Geometridae, which had two caterpillars. The time-length varied among herbivores because of the different rate of plant consumption. While for Monoplatus sp. and E. meditabunda the experiment took 7 days, for the voracious Geometridae caterpillars the plants were collected after 2 days. For the drought treatment the plants were kept for 7 days until the soil became dry while in the case of the high temperature treatment, the plants were maintained at 40°C for 7 days. For the dark treatment, the plants had the photoperiod turned off for 7 days or until the plants showed a sign of a loss of chlorophyll (4–5 days). After each treatment using five replicates, the plants were individually extracted, and the crude CHCl₃ fractions were submitted to ¹H NMR analysis. The resulting data was then subjected to a multivariate analysis.

The leaves of the mature plants (approximately 6 months old, when plants start to produce seeds) and seedlings (1–4 months old) were frozen immediately after sampling using liquid N₂ and crushed in a mortar and pestle until a fine powder was obtained. A total of 100 mg of ground plant material was then transferred to a 2 mL Eppendorf tube and vortexed for 5 min with a mixture of solvents (1 mL; chloroform:methanol:water; 2:1:1; v/v/v). The samples were then centrifuged at 10,000 rpm at 0°C for 10 min, and organic phases were collected using a pipette. The extraction procedure was repeated, and the pooled chloroform fraction was dried under a N₂ stream, while the aqueous extract was dried using a SpeedVac (vacuum centrifuge). The resulting samples were stored at −20°C until analysis.

Fresh aerial parts of P. pellucida (500 g) obtained from cultivated plants were extracted using the same procedure as above and after concentration using a rotary vaporator, 4.12 g of crude CHCl₃ fraction was obtained. This extract was suspended in MeOH:H₂O (1:4, v/v, 400 mL) and subjected to a dechlorophylation step in a Celite column as described (Fernandes et al., 1997), yielding a chlorophyll-free fraction (1.12 g). This fraction (1.00 g) was then separated with a VLC system using silica gel, eluted with a gradient of hexanes:EtOAc (9:1 – 0:1) and then with EtOAc:MeOH (9:1 – 0:1), yielding 95 fractions (10 mL each). Similar sub-fractions were pooled (up to 15 fractions) based on TLC analysis. The main compounds in the fractions were purified by circular chromatography (Chromatotron^®) eluted with a gradient of hexanes:EtOAc (9:1 – 0:1) yielding 2,4,5-trimethoxycinnamic acid (1, 6.0 mg), 2,4,5-trimethoxystyrene (2, 14.1 mg), 2,4,5-trimethoxybenzaldehyde (3, 19.7 mg), dillapiol (4, 45.8 mg), pellucidin A (5, 3.4 mg), sesamin (6, 15.3 mg), and 5,6,7-trimethoxyflavone (7, 121.3 mg).

Pellucidin A (5), 5,6,7-trimethoxyflavone (7), sesamin (6), 2,4,5-trimethoxystyrene (2), 2,4,5-trimethoxycinnamic acid (1), 2,4,5-trimethoxybenzaldehyde (3), and dillapiol (4) were quantified by HPLC-DAD. Calibration curves were generated using standard solutions of the respective purified compounds. A stock solution in MeOH (1 mg/mL) was prepared for each compound and diluted to 1.000, 0.5000, 0.2500, 0.125, 0.0625 and 0.0312 mg/mL. All of the solutions were analyzed by HPLC under the same conditions of analysis as the extracts.

Despite the diverse reports of natural compounds containing cyclobutane rings, some authors claim that these compounds could be formed during the extraction process (Seidel et al., 2000; McCracken et al., 2012). Thus, the dimerization of 2,4,5-trimethoxystyrene to pellucidin A was evaluated during the extraction procedures of leaves of P. pellucida. In two sets of extractions, 1 mg of ¹³C-natural abundance 2,4,5-trimethoxystyrene or (8-¹³C)-2,4,5-trimethoxystyrene solutions in CHCl₃ (500 μL) were spiked in the extraction procedures, using the protocol for metabolome analysis. The formation of pellucidin A under such conditions was evaluated by comparison of relative intensities of molecular ions by GC-MS analysis, in which the abundance of [M+1]^+. and [M+2]^+. provide data for incorporation of one or two unities of ¹³C-labeled precursors (Opitz et al., 2014). The data were compared with the control experiments without spiking either natural or labeled 2,4,5-trimethoxystyrenes.

An amount of 5–7 mg of the crude extracts dissolved in CDCl₃ was submitted to ¹H NMR analysis at 300 MHz, and the data were processed using the program MestreNova. The scale was adjusted with the TMS signal as 0.00 ppm, the range of chemical shift data was between 1.4 and 12 ppm, and the residual chloroform signal between 7.20 and 7.28 ppm was discarded. A binning was made for a range of 0.02 ppm, and the data were normalized by the total area. The data were saved as an ASCII text file and processed by The Unscrambler program (version 9.5, CAMO Process AS, Norway).

To investigate the substrate specificity in the formation of pellucidin A in P. pellucida, a series of synthetic ¹³C-labeled cinnamic acids and styrenes were evaluated by in vivo feedings and enzymatic conversion experiments. The cinnamic acids were synthesized by a Knoevenagel condensation between malonic acid and benzaldehyde derivatives. Malonic acid (9.60 mmol) and the benzaldehydes (10.00 mmol) were dissolved in pyridine (3.5 mL), followed by addition of 10 μL of piperidine and 10 μL of aniline, both previously distilled, as described (Katayama et al., 1992). After the reaction was completed, the mixture submitted to work-out yielded colorless crystals of cinnamic acids (Supplementary Figure 9; A1–A16) with a typical yield above 90%. The structures of the cinnamic acids were characterized using ¹H and ¹³C NMR and EIMS data in addition to a comparison with previously published data (Katayama et al., 1992; Mitra and Karchaudhuri, 1999). For the preparation of isotopically labeled versions of cinnamic acids, (2-¹³C)-malonic acid (99%) from Sigma-Aldrich was used and the corresponding cinnamic acids were purified and characterized spectroscopically.

Into a 50 mL two-necked flask under reflux was added 125 μL (2 mmol) of methyl iodide and 0.84 g (3.2 mmol) of triphenylphosphine in 30 mL of dry tetrahydrofuran. The reaction was kept for 12 h at a temperature of 110°C. Then, the white solid formed was washed with hexane (3 × 50 mL) and vacuum dried (Sato et al., 2005).

The styrenes were prepared via the Wittig reaction using the respective benzaldehydes and the methyl triphenylphosphonium iodide prepared above (Simpson et al., 2005; Farina et al., 2007). Then, 3 mL of dry tetrahydrofuran and 1.2 mmol methyltriphenylphosphonium iodide were added into a 25 ml three-necked flask equipped with magnetic stirring. The mixture was stirred at −40°C, and after 20 min, 1 mL (12.85 mmol) of n-BuLi was added, after which the slightly yellow solution was kept in an ice bath at 0°C under stirring for 30 min. Following that, a solution of 1.2 mmol of benzaldehydes in 2 ml of dry THF was added over approximately 30 min and stirred for 10 h at room temperature. The solution was then poured into 50 mL of ice water and extracted with hexanes (3 × 50 mL), dried and concentrated to dryness. The crude extract was then subjected to a 4 cm flash silica column using hexanes:EtOAc (9:1; v/v) as the mobile phase, yielding the styrenes (Supplementary Figure 10; S1–S16) and two hydrogenated derivatives (S17 and S18). For the preparation of the isotopically labeled styrenes, the methyl triphenylphosphonium iodide was prepared from ¹³CH₃I.

The preparation of the phenolic styrenes was performed starting from the demethylation of the respective methoxy-benzaldehydes using BBr₃ (Unver et al., 2011). To a solution of 1.2 mmol of the methoxy benzaldehydes in chloroform (12.5 mL) was added 0.306 mmol (2 eq) BBr₃ at 0°C. The reaction mixtures were stirred under a N₂ stream for 12 h. After that, a solution of 25 mL of sodium hydroxide was added. Then, concentrated H₂SO₄ was added, and the precipitate was extracted with ethyl ether. The organic fraction was dried by anhydrous Na₂SO₄, filtered, and concentrated under vacuum using a rotary evaporator.

Hydroxybenzaldehydes (2.40 mmol), tert-butyldimethylsilane chloride (2.88 mmol), and imidazole (4.8 mmol) were added to a 50 mL round bottom flask and were heated in a microwave at 90 W (2 min) and then to 180 W (3 min). The reaction was quenched by the addition of 15 mL of water, followed by extraction with ethyl ether. The organic phases were combined, dried with anhydrous Na₂SO₄ and evaporated under reduced pressure. The crude product was purified in a short VLC column eluted with hexanes:EtOAct (7:3; v/v), with typical yield of the reactions being 80%. All protected compounds were characterized by GC-MS analysis (Bastos et al., 2005).

The TBS-protected phenolic styrenes (0.898 mmol) were dissolved in 3.0 mL of dry THF. The solution was cooled to 0°C in an ice bath while constantly being stirred for 30 min. Then, a solution of 312 μL of tetrabutylammonium fluoride (TBAF) was added slowly to 1.2 mol/L in dry THF, while the mixture was stirred for 30 min. The crude products were purified by VLC eluted with hexanes:EtOAct (4:1; v/v), and a total of 16 styrenes were subsequently isolated and characterized by GC-MS and ¹H and ¹³C NMR analysis. The hydrogenated derivatives (S17 and S18) were obtained by treatment of corresponding styrenes (S1 and S10) with H₂-Pd/C in dichloromethane at room temperature overnight.

The leaves of P. pellucida (3–4 months old) were cultivated for 21 days in a hydroponic system. After this period, the leaves with root formation were transferred to Eppendorf tubes containing a solution of isotopically labeled precursors in pure water and kept for various periods of time. Specific details for the incorporation of each substrate are provided in the following sections.

Solutions containing 1.5, 3.0, and 6.0 μmol of L-(2-¹³C)-phenylalanine in 100 μL of Milli-Q water were used. Aliquots of 1.0 mL of the different concentrations of the isotopically labeled precursors were administered in the hydroponic system of the plants maintained in Eppendorf tubes. The administration periods of the isotopically labeled precursors in the plants were 12, 24, 48, 72, 96, and 120 h at 25°C, with replenishment of the precursor solutions when needed. All experiments were carried out in triplicate and the blank control was administered water with DMSO (0.5%) with no labeled precursor added.

The leaf samples were then frozen in liquid nitrogen and extracted with CHCl₃:MeOH:H₂O (2:1:1; v/v/v). The chloroform fraction was separated, and the solvent was evaporated under vacuum. The samples were resuspended in MeOH filtered through a 0.45 μm membrane and subjected to GC-MS analysis. The quantification of the precursor incorporations was performed as described (Opitz et al., 2014).

The administration of (8-¹³C)-cinnamic acids (A1–A16) and (8-¹³C)-styrenes (S1–S18) in plantlets of P. pellucida was performed under the same conditions as for L-(8-¹³C)-phenylalanine. The (8-¹³C)-cinnamic acids were solubilized in 0.5 μl of DMSO and brought to 1 mL with water. Each of the concentrations, 1.0, 2.1, and 4.2 μmol of the cinnamic acids in 100 μL of the solution, were fed for the same intervals as in the above section. The plant material was then treated, the crude extracts analyzed by GC-MS, and the incorporation of the precursors was quantified as above (Opitz et al., 2014).

Leaves crushed using liquid N₂ were extracted with 100 mL of pH 7.0 potassium phosphate buffer and EDTA (0.12 mmol), ascorbic acid (0.2 mmol), MgCl₂ (0.31 mmol), and polyvinylpolypyrrolidone (45.0 mmol) under gentle stirring for 30 min at 4°C followed by filtration with miracloth tissue. An aliquot of the resulting supernatant was separated, and the precipitate was discarded. Saline precipitation was performed using the soluble fraction from the first centrifugation, slowly adding 1.0 mmol of (NH₄)₂SO₄, under constant stirring. The solution was centrifuged for 10 min at 13,000 rpm, yielding a pellet that was used as the protein source for the enzymatic assays. The protein contents were determined as 33.3 ± 3.7 mg/mL (Bradford, 1976). The enzymatic conversions were performed using 100 μL of protein extract, 20 μL of the substrate solutions at 10 mM and 76 μL of potassium phosphate buffer at 40 mmol.L^–1 (pH 7.0), as described (Vassao et al., 2006). The reaction mixtures were incubated for 30, 60, 90, or 120 min at 25°C, after which they were terminated by extraction with EtOAc (2 × 500 μL). The blank control had only substrates without proteins or only protein preparations using the same incubation buffer. To verify a possible oxidative coupling reaction, H₂O₂ and 50 mM NADPH were tested as oxidants both together and separately (Lu et al., 2004; Vassao et al., 2006). The enzymatic fractions, precursors and crude reaction products were analyzed by HPLC, using a Phenomenex^® reverse phase C-18 column (Luna C-18 150 × 4.6 mm, 5 μm), with a flow of 1 mL.min^–1, and a detector set at λ = 260 nm.

To elucidate the biosynthetic pathway of pellucidin A, preliminary feeding experiments were performed with L-(2-¹³C)-phenylalanine. The percentage of incorporation was determined by GC-MS analyses of the crude extracts of P. pellucida leaves incubated with isotopically labeled precursors compared to the control leaf extracts and pure pellucidin A (Tables 2, 3).

First, considering the biosynthetic possibilities for pellucidin A formation (Figure 4), accurate analysis of the mass spectra of putative intermediates after feeding L-(2-¹³C)-phenylalanine indicated significant enrichment of 1.49 and 3.92% to 2,4,5-trimethoxycinnamic acid and 2,4,5-trimethoxystyrene (Table 3, Figure 5, and Supplementary Figure 5; 96 h), respectively. Then, the analysis of the mass spectrum of pellucidin A revealed the level of incorporation. For instance, the mass spectrum of pellucidin A at natural abundance displayed the molecular ion at m/z 388, and the fragment ions at m/z 194.0 and 179.0 Da with naturally abundant ions with ¹³C being compatible with intensities in the range of 0.07% (Figure 5 and Supplementary Figure 35). The feeding with L-(2-¹³C)-phenylalanine led to an increase in the percentage of ions at m/z 389 [M+1]^+., 195, 180, and 152 assigned to ¹³C-enriched pellucidin A with incorporation of one unit of L-(2-¹³C)-phenylalanine while the increasing percentage of the masses at 390 [M+2]^+. was attributed to pellucidin A enriched with two molecules of L-(2-¹³C)-phenylalanine (Figure 5). The level of incorporation of L-(2-¹³C)-phenylalanine in the time course experiments after 96 h of incubation led to a maximum of 0.72% enrichment in pellucidin A (Table 3).

The GC-MS analysis of crude extracts of P. pellucida leaves incubated with (8-¹³C)-cinnamic acid and (8-¹³C)-ferulic acid showed the incorporation of these precursors into 2,4,5-trimethoxystyrene and pellucidin A. First, the incorporation of (8-¹³C)-cinnamic acid over the time course of 12–120 h resulted in the incorporation into 2,4,5-trimethoxystyrene and pellucidin A of a maximum of 2.89 and 1.32%, respectively (Table 3 and Supplementary Figure 5; 96 h). Then, the incubation of P. pellucida with (8-¹³C)-ferulic acid for 48 h led to its incorporation into both 2,4,5-trimethoxystyrene (0.69%) and pellucidin A (0.51%) (Table 3 and Supplementary Figure 6). The percentages of incorporation are in the range of incorporation of L-(2-¹³C)-phenylalanine into pellucidin A (Supplementary Figure 4) and are meaningful considering the number of steps required for L-phenylalanine to be initially converted to ferulic acid, and the additional further steps required to produce pellucidin A. Nevertheless, plants incubated for a period of time longer than 48 h showed signs of browning as compared with other feeding experiments and precluded further label chasing. The phenolic nature and the low solubility of (8-¹³C)-ferulic acid in aqueous solution may have contributed to difficulties in transportation to the biosynthetic site(s), as previously demonstrated (Davin et al., 2003; Vassao et al., 2006). In fact, several biosynthetic studies have shown low levels of cinnamic acid and ferulic acid incorporation, such as in the case of lignan podophyllotoxin in young plants (2–3 years) of Podophyllum hexandrum, with uptake of 0.04 and 0.053%, respectively (Jackson and Dewick, 1984). The incorporation of cinnamic acids in the (+)-lioniresinol found in Lyonia ovalifolia was in the range of 0.013–2.75% (Rahman et al., 2007). In the case of lignans such as (−)-cis-blechnic, (−)-trans-blechnic acids, and (−)-brainic acids occurring in the fern Blechnum spicant, the percentages of incorporation were significantly higher between 1.0 and 4.2% (Davin et al., 2003). The incorporation of the cinnamic acids is also described in studies of norlignan biosynthesis of (Z)-hinokiresinol in Asparagus officinalis (Suzuki et al., 2001) and agatharesinol from Cryptomeria japonica and Agathis australis (Imai et al., 2006). The individual administration of ¹³C-enriched cinnamic acid demonstrates that the side chain ¹³C-7, ¹³C-8, and ¹³C-9 atoms of cinnamic acid are incorporated into C-1 and C-3, C-2 and C-4, and C-5 of (Z)-hinokiresinol with variable percentages between 0.56 and 32.3%.

The GC-MS analyses of the crude extracts from P. pellucida leaves fed with (8-¹³C)-2,4,5-trimethoxycinnamic acid showed its sequential incorporation into 2,4,5-trimethoxystyrene and pellucidin A, reaching a maximum of 9.0 and 7.5%, respectively (Table 3 and Supplementary Figure 7; 96 h). The comparison between the relative abundances of the molecular ion of pellucidin A (m/z 388 Da, 0.34%; m/z 389 Da, [M+1]^+., 0.07%; m/z 390 Da, [M+2]^+., 0.03%) and the enriched version, revealed a higher relative intensity for [M+1]^+. (m/z 389 Da, 0.27%) and [M+2]^+. (m/z 390 Da, 0.04%) (Table 2 and Figures 4, 5).

Having characterized the in vivo incorporation of (8-¹³C)-2,4,5-trimethoxycinnamic acid to pellucidin A, experiments using enzymatic conversion were further conducted to evaluate the conversion of a series of unlabeled 2,4,5-trimethoxystyrene and 2,4,5-trihydroxystyrenes. The assays with enzymatic fractions from leaves of P. pellucida were carried out at varying temperatures, pH values and reaction times (data not shown). The extracts resulting from the assays were analyzed by HPLC to characterize the conversion of precursors to pellucidin A or related compounds. In fact, the comparison of the HPLC analyses of the control (enzyme extract) (Supplementary Figure 8) with the enzymatic fractions and the respective standards characterized the conversion of 2,4,5-trimethoxycinnamic acid to 2,4,5-trimethoxystyrene as the main product of the enzymatic conversion (Figure 4).

Further enzymatic conversion experiments were conducted to determine the level of specificity for the decarboxylation reaction. Thus, various cinnamic acids were tested to evaluate the formation of the respective styrenes including 2,4,5-trimethoxycinnamic acid, 3,4,5-trimethoxycinnamic acid, and 3,4-dimethoxycinnamic acid. Such reactions were observed in the first 30 min, after which the formation of the products remained practically constant up to 120 min without a significant increase in the yield (data not shown). Therefore, the decarboxylation of cinnamic acids in P. pellucida has apparently no specificity associated to the substitution pattern of the aromatic ring of cinnamic acids, although no compounds having electron withdrawing substituents at conjugated positions were evaluated. Enzymes capable of carrying out decarboxylation reactions have been described, as in the case of the ferulic acid decarboxylase from Bacillus pumilus and Enterobacter sp. (Gu et al., 2011), as well as p-coumaric acid decarboxylase purified from Lactobacillus plantarum (Cavin et al., 1997). The presence of these enzymes in plants is reported for species of Catharanthus roseus, Nicotiana tabacum, and Daucus carota (Takemoto and Achiwa, 1999). The action of decarboxylases in the biosynthesis of norlignans (E)- and (Z)-hinokiresinol in A. officinalis was also observed, in which the 4-coumaroyl 4-coumarate underwent a decarboxylation step followed by the formation of a bond between the carbons C7–C8′ leading to hinokiresinols (Suzuki et al., 2002; Suzuki and Umezawa, 2007).

The GC-MS analyses of the crude extracts of P. pellucida leaves resulting from incubation with (8-¹³C)-2,4,5-trimethoxystyrene showed its incorporation into pellucidin A with an incorporation of 12.8% at 96 h (Figure 5, Table 3, and Supplementary Figure 9). The incorporation of two molecules is noticeable with the relative intensity of the [M+2]^+. of labeled pellucidin A being near 0.17% of relative abundance, contrasting with 0.05 and 0.08% to the molecular ion and to [M+1]^+. ion of the natural version, respectively. In fact, the overall analysis of double-labeled pellucidin A, indicate the dilution effect of using L-phenylalanine, which is a primary precursor for coniferyl alcohol required for the biosynthesis of lignin, while for feeding labeled downstream intermediates in the formation of pellucidin A, the level of incorporations increases sharply. Thus, the formation of pellucidin A is confirmed to have 2,4,5-trimethoxystyrene as a direct precursor, and its formation should occur through a dimerization reaction between two 2,4,5-trimethoxystyrene units (Figure 4). This result contrasts with the biosynthetic suggestion for pellucidin A that precludes the possibility of dimerization between two 2,4,5-trimethoxystyrene units due to steric hindrance caused by the methoxy groups at positions 2/2′ of the aromatic rings (Bayma et al., 2000). However, as discussed above, pellucidin A has a trans configuration between the aromatic ring in the cyclobutane ring, which would minimize the repulsion between the methoxy groups at the 2/2′ positions.

According to the biosynthetic hypothesis, the incorporation of (8-¹³C)-2,4,5-trihydroxystyrene (Supplementary Figure 10; S10) would lead to the formation of a cyclobutane dimer (m/z 304 Da), followed by a series of methylations to produce pellucidin A (Bayma et al., 2000). The analysis of the GC-MS of the extracts resulting from the incubations did not detect the putative dimer produced from dimerization of (8-¹³C)-2,4,5-trihydroxystyrene. Feeding with the alternative labeled substrates (8-¹³C)-2,4,5-trihydroxyethylbenzene (S17), (8-¹³C)-3,4-dihydroxystyrene (S11) and (8-¹³C)-4-hydroxystyrene (S12) to determine whether methylations in the phenolic hydroxyl groups would take place as intermediate steps leading to pellucidin A (Figure 4) was also, not supported by GC-MS analysis of the crude extracts precluding such possibilities.

In addition to the ¹³C-cinnamic acids and ¹³C-styrenes used in the feeding experiments, unlabeled 3,4-dimethoxy- and 2,4,5-trimethoxypropenylbenzenes (Supplementary Figure 10) were also evaluated as substrates in enzymatic conversion using leaves of P. pellucida. The objective was to determine whether the enzymes from leaves of P. pellucida can produce pellucidin A analogs or related dimers, such as magnosalin or endiandrin A, dimers described from Endiandra anthropophagorum and Piper cubeba, respectively (Badheka et al., 1987; Davis et al., 2007). The analysis of extracts from P. pellucida resulting from incubation using enzymes with the different cinnamic acids and styrenes by GC-MS did not show the production of any other compound containing a cyclobutane ring.

Several examples of cyclobutane dimers of natural products including those resulting from pyrones and chalcone have been suggested as artifacts formed during compound isolation (Seidel et al., 2000; McCracken et al., 2012). Considering that this possibility could also be applied to the case of pellucidin A in P. pellucida, its formation during the extraction was investigated using the (8-¹³C)-2,4,5-trihydroxystyrene spiked in the solutions (see experimental). The analysis of relative intensities of the molecular ion peak, compared to the expected contribution of [M+1]^+. and [M+2]^+., did not support the formation of pellucidin A as an artifact. Nevertheless, the in vivo irradiation of plants under UV₃₆₅ light led to an almost 200% increase in the content of pellucidin A (Supplementary Figure 2). Additional compounds that could be produced from dimerization of styrenes are pachypostaudin B, which has already been isolated from the extracts of P. pellucida (Kartika et al., 2020). However, pachypostaudins A and B are found as racemates in the extracts of P. staudtii bark, and were also obtained by thermal dimerization of 2,4,5-trimethoxystyrene (Ngadjui et al., 1989). The formation of pellucidin A could also be considered a photochemical artifact, but pachypostaudins were not detected in our study even in low amounts similar to pellucidin A in P. pellucida. Therefore, the biosynthesis of pellucidin A results from a more complex mechanism which requires studies of the molecular aspects, specificity for substrates, localization of biosynthetic steps and of its physiological role for the plant fitness.

Taken together these results indicate that the formation of pellucidin A in P. pellucida can be formed naturally having the L-phenylalanine followed by a series of steps of deamination forming cinnamic acid, then oxidation by phenolase, O-methylations leading to 2,4,5-trimethoxycinnamic acid, decarboxylation and then by a cycloaddition [2+2] reaction of 2,4,5-trimethoxystyrene to produce pellucidin A.