Two red clover (T. pratense L.) populations, cv. Milvus (wild-type, Aa4381) and a low leaf PPO mutant line (Aa4512; Winters et al., 2008), were sown in John Innes no. 3 compost. The mutant population was a genetic mutant for low leaf PPO, selected from and backcrossed into cv. Milvus (Lee et al., 2004). Leaf PPO activity was confirmed as described elsewhere (Winters et al., 2008).

When plants were grown in a glasshouse, growth conditions were set to a minimum 12 h photoperiod, minimum light intensity of 300 μmol m⁻² sec⁻¹, and day/night temperature of 20/16°C. Plants (8 weeks old) that were acclimatized to a controlled environment in a growth cabinet were maintained under a 16 h photoperiod, 400 μmol m⁻² sec⁻¹ and 20/16°C for 4 weeks prior to treatment.

Seeds of wild-type and low leaf PPO mutant plants were sown 4 per pot) and grown in a glasshouse. At 3 weeks, measurements of germination rate, number of shoots per plant, and number of leaves per shoot were taken (n = 60) before plants were reduced to 1 plant per pot. Plant height was recorded at intervals up to 8 weeks after sowing (n = 15). Height measurements were taken from the base of the plant to the node of the uppermost leaf on the longest shoot. Dry matter yield of leaf and stem material was measured separately and whole leaf samples were flash frozen and stored at −80°C or freeze-dried and milled for isoflavonoid and hydroxycinnamate analysis.

For a comparison and analysis of isoflavonoids and hydroxycinnamates in whole leaves, leaf samples were taken from glasshouse grown mature, 8 week old wild-type and mutant (low leaf PPO) red clover plants. These were freeze-dried, milled, and weighed (50 ± 2.5 mg samples) and then extracted for isoflavonoids and hydroxycinnamic acids as described below.

Equivalent plants were acclimatized to a controlled environment (16 h photoperiod, 400 μmol m⁻² sec⁻¹, 20/16°C) for 4 weeks and used for a more detailed analysis of the effect of light on phaselic acid (caffeoyl malate) and coumaroyl malate in isolated chloroplasts from low leaf PPO mutant and wild-type plants. Randomly selected leaves were sampled separately (n = 6) for whole leaf extraction (6 × 1 g samples) and chloroplast purification (6 × 10 g samples). Samples were harvested from six 12 week old plants in a two phase cross over design; plants were either illuminated or kept in dark for a minimum of 7 and 16 h, respectively. After the first sampling in the light or dark, the same plants were transferred to the opposite conditions (i.e., light to dark and dark to light). The remaining leaves were harvested after the equivalent time in the light or dark. Thus, 2 × 3 light adapted plants were sampled 7 h into a light period while 2 × 3 dark adapted samples were harvested after 16 h darkness at room temperature; the difference between plants was the level of stress caused by removal of leaves for harvest at the first sampling.

Chloroplasts were isolated from leaves (6 × 10 g samples) essentially as described by Mills and Joy (1980) except that initial homogenization was performed with a hand-held blender (Stabmixer SM3, Primo, Antwerp, Belgium) for three cycles of 5 s, with a 10 s pause between each. Intact chloroplasts were collected by centrifugation through 40% Percoll, the pellet resupended in <1.0 ml buffer A (Mills and Joy, 1980) and intactness confirmed by microscopy under UV illumination (Olympus BH2-RFCA T3; Olympus, Tokyo, Japan). The remainder was centrifuged for 1 min at 4°C and 2,500 × g to remove any remaining Percoll and the pellet was resuspended in 300 μL pre-cooled lysis buffer C.

Isoflavonoids and hydroxycinnamic acids were extracted from the three sample types (freeze-dried leaf samples, whole leaf samples, and lysed chloroplast suspension) in methanol. The stability of biochanin and formononetin at 90°C for 5 min has previously been demonstrated (Stintzing et al., 2006). The freeze-dried leaf samples (50 ± 2.5 mg samples) were extracted x1 in 0. 5 mL 70% methanol at 90°C for 5 min followed by x2 in 0.5 mL 70% methanol at room temperature. The whole leaf (1 g) samples were ground in liquid nitrogen pre-cooled pestle and mortars to a fine powder to which 10 mL of 70% methanol at 90°C was added for 5 min. Lysed chloroplast samples were brought to a final concentration of 70% methanol by addition of 100% methanol. Aliquots of methanolic extracts from chloroplast and whole leaf samples were taken to assess chlorophyll content according to Arnon (1949). The remainder of the samples were treated to decrease the methanol concentration to <10% by either rotary evaporation or dilution with deionized water. These were then applied to pre-conditioned (4 mL of 100% methanol) and equilibrated (4 mL of 5% acetic acid) Waters Sep-Pak Vac RC (500 mg) C₁₈ reverse-phase extraction cartridges (Waters, Milford, USA) on a vacuum manifold. After washing with 5% acetic acid, the lower-polarity analytes of interest were eluted with 100% methanol. These fractions were evaporated to complete dryness, the residue re-dissolved in 80 μL of 100% methanol and insoluble material removed by centrifugation at 14,100 × g for 4 min. The supernatant was retained for metabolite profiling by HPLC-PDA-ESI-MSⁿ.

HPLC/MSⁿ analysis of chloroplast and total leaf extracts was carried out with a Thermo Finnigan LTQ system (Thermo Electron Corporation, USA). The system was fitted with a Waters C18 reverse phase Nova-Pak column (4 μM, 3.9 × 150 mm; Waters, Milford, USA) which was maintained at 30°C. Injection volumes of 10 μL were separated by HPLC and the UV absorbance was recorded (240–400 nm). The mobile phase consisted of 0.1% formic acid in deionized water (mobile phase A) and 0.1% formic acid in methanol (mobile phase B). The percentage of mobile phase B was increased linearly from 0 to 65% over 45 or 60 min, with 0.1 mL min⁻¹ going to the mass spectrometer. Mass spectra were obtained using a Finnigan LTQ linear ion trap instrument with an ESI source (Thermo Electron Corporation, USA). Nitrogen was used as the sheath (40 units) and auxiliary gas (25 units) and helium was used as the collision gas. Spectra were obtained in negative ionization mode with the following settings: spray voltage, 4 kV; capillary temperature, 320°C; capillary voltage, −45 V; tube lens, −93.1 V. In positive ionization mode the settings were as follows: spray voltage, 4 kV; capillary temperature, 320°C; capillary voltage, 44 V; tube lens, 85 V. Generally collision energy of 35% was used, however, where limited ion fragmentation was observed collision energy was raised to 70%.

Compounds were identified with based on the characteristic MSⁿ fragmentation patterns of isoflavonoid (daidzein, genistein, biochanin A, and formononetin) and hydroxycinnamic acidse (caffeic acid, p-coumaric acid, and ferulic acid) standards. Some peaks eluting close together were observed to have the same mass and fragmentation pattern and were therefore designated with the same ID number. This may be due to the separation of the dissociated and non-dissociated forms of acids. Metabolite abundance was determined by integration of peak area (Qual Browser of the Xcalibur 1.4 software; Thermo Electron Corporation, USA). These measurements were then expressed on the basis of the chlorophyll content in order to predict the likelihood of chloroplastic subcellular localization. The compound was viewed as chloroplastic and not due to contamination if ≥0.1% of the whole leaf concentration was detected within chloroplasts. If the percentage was ≤0.01%, then the compound was assumed to accumulate mainly outside of the chloroplast.

Significant differences between the growth parameters of the two varieties were analyzed by repeated measurements by REML; other growth and metabolite data were analyzed using the student T-test and ANOVA, GenStat 14th Edition.

Known PPO substrates belong to just two classes of polyphenols: flavonoids and phenolic acids (Parveen et al., 2010). Flavonoids have a fifteen carbon structure, comprising two aromatic rings (A- and B-rings) linked by a three carbon bridge (C-ring; Figure 3). Subclasses of flavonoids (isoflavonoids, flavones, flavonols, anthocyanins, and flavanols) are known to include some PPO substrates however, an even larger number are hydroxycinnamates, a subclass of phenolic acids with a characteristic C₆-C₃ structural skeleton (Figure 4); these include the known red clover PPO substrates phaselic acid (caffeoyl malate) and caffeoyl DOPA (clovamide) (Sullivan and Hatfield, 2006; Winters et al., 2008). A comprehensive tandem MS analysis was performed in order to identify new red clover PPO substrate candidates located in leaves and chloroplasts (Figure 5).

PPO can utilize flavonoids as substrates where one or more hydroxyl groups are present on the B ring (Martinez and Whitaker, 1995; Parveen et al., 2010). In this study isoflavonoids were identified as both the hexoside conjugated and unconjugated forms of formononetin and biochanin A, irilone hexoside, and 2-hydroxy-2,3-dihydrogenistein hexoside (see Figure 5 and Table 3). As the functional groups on the B ring of the major isoflavonoids detected (biochanin A and formononetin) are restricted to a single methoxy group, and there is no evidence that red clover leaf PPO enzymes can catalyze oxidation of unconjugated formononetin, biochanin A, or genistein it is considered unlikely that the isoflavonoids identified in this study are endogenous red clover PPO substrates.

The hydroxycinnamic acids, phaselic acid (caffeoyl malate; peak 3) and caffeoyl L-3,4-dihydroxyphenylalanine (N-caffeoyl L-DOPA or clovamide; peak 5) are the main PPO substrates in red clover leaves (Sullivan and Hatfield, 2006; Winters et al., 2008) and were clearly identified in this study (Figure 5 and Table 3). In addition caffeoyl conjugates, caffeoyl mevalonate (peak 9, Figure 5A) and caffeoyl tyrosine (peak 7a, Figure 5A) were also tentatively identified and these compounds have very similar structural characteristics to phaselic acid and caffeoyl DOPA respectively (Figure 4). Hence caffeoyl mevalonate and caffeoyl tyrosine are potential endogenous substrates for red clover PPO (Parveen et al., 2010). In addition to the caffeoyl esters and amides, feruloyl DOPA (peak 8; Table 3) was also observed. Because red clover PPOs have also been reported to oxidize the o-diphenol DOPA (Parveen et al., 2008; Winters et al., 2008), feruloyl DOPA (peak 8) is another endogenous red clover PPO substrate candidate.

The hydroxycinnamic acid structures thus far discussed are typical of candidate PPO substrates because of their o-diphenolic structure with readily oxidizable OH-groups (Martinez and Whitaker, 1995; Parveen et al., 2010). The o-diphenolic or catecholase activity is characteristic for PPOs and is the catalytic activity most frequently discussed in the literature. However, some PPOs also have monophenolase activity (Steffens et al., 1994). This has been demonstrated for red clover leaf extracts using monophenolic p-coumaric acid and a catalytic amount of o-diphenol (Winters et al., 2003). p-coumaric acid conjugated with an organic acid (peak 6, coumaroyl malate), a glycoside (coumaroyl hexoside), and an amino acid conjugate (peak 10, coumaroyl tyrosine) were observed in our extracts (Figure 5 and Tables 3, 4).

An early report by Satô (1966) showed a copper-containing enzyme catalyzing the oxidation of p-coumaric acid to caffeic acid in isolated chloroplasts of several plant species including Saxifraga stolonifera; this was subsequently found to express PPO enzyme activity (Jeong et al., 2005). Moreover, in a study with chloroplasts isolated from Beta vulgaris, Halliwell (1975) observed that hydroxylation of p-coumaric acid only occurred in the light unless a reductant such as ascorbate, NADH, or NADPH was included. On the basis of these observations they proposed that this reaction is catalyzed by PPO monophenolase activity which is dependent on a reducing environment. Thus, the identification of coumaroyl esters in red clover leaf extracts could be indicative of novel, potentially chloroplastic, monophenolic PPO substrate candidates, and a precursor for phaselic acid.

In order to determine the likelihood that identified candidates are PPO substrates in intact tissue, the hydroxycinnamate profile of whole leaf extracts and purified chloroplasts was compared (Figures 5A,B). The profile obtained from the chloroplast extracts was distinct from that from whole leaf extracts with six isoflavonoids and two potential PPO substrates (coumaroyl hexoside and coumaroyl malate) showing enrichment in the plastid fraction (Table 4).

Chloroplast preparations were examined microscopically prior to lysis to confirm that the chloroplasts were mainly intact. However, we acknowledge that the integrity of the membrane could be breached allowing bidirectional flow of constituents between chloroplast and cytosol. However, comparison of chromatograms from chloroplast and whole leaf extracts showed that a major peak corresponding to phaselic acid in total leaf extracts (RT 17.1 min) was only just detectable in the chloroplast profile (Figure 5B, and Table 3). As phaselic acid has been previously shown to be present in leaves of red clover (Sullivan and Hatfield, 2006; Winters et al., 2008; Sullivan, 2009) this indicates that the lysed chloroplast preparations were free from contamination and could be used to identify candidate PPO substrates specifically targeted to the chloroplasts.

Coumaroyl malate, monophenolic PPO substrate candidate, appeared to be enriched in red clover chloroplasts (Figures 5A,B and Table 4), along with coumaroyl hexoside (Figure 6 and Table 4). By contrast, the known o-diphenolic red clover PPO substrates phaselic acid and clovamide are predominantly located in extra-chloroplastic compartments, with on average only 0.002 and 0.0004%, respectively of the total leaf concentration being detected (Table 4) in chloroplasts isolated from unstressed leaves. In addition, other candidate substrates, such as caffeoyl tyrosine (peak 7a), feruloyl DOPA (peak 8), and caffeoyl mevalonate (peak 9) were all below detection limits in red clover chloroplasts (Table 4). Based on these data, caffeoyl derivatives that would be generated from PPO catalyzed oxidation of coumaroyl conjugates do not accumulate in chloroplasts but are exported to the cytoplasm. Since PPO is generally present in a latent inactive form, it is possible that evidence of such activity can only be observed during periods of stress.

Purified red clover chloroplast extracts were particularly abundant in isoflavonoids (Table 4), most notably the aglycone forms of formononetin and biochanin-A. This observation is in agreement with the literature which documents the presence of flavonoids in the chloroplast of the Atlas 68 barley variety (Saunders and McClure, 1976b), the green olive tree (Phillyrea latifolia; Agati et al., 2007), tea (C. sinensis; Liu et al., 2009) and in 23 out of 25 other vascular plant species surveyed (Saunders and McClure, 1976a). The role of flavonoids in chloroplasts remains unclear, although they may be required for the protection of the plant's active photosynthetic machinery. Barley plastids selectively accumulated the C-glycosyl flavone, saponarin, and since this accumulation is far-red reversible, it is possibly controlled by phytochrome (Saunders and McClure, 1976b). Agati et al. (2007) found that photo-induced generation of ¹O₂ was inversely related to the content of flavonoids in the mesophyll cells of P. latifolia leaves. These data suggest a complementary protective role for flavonoids in photosynthetically active chloroplasts under high light stress.

The presence of PPO in the chloroplast has long prompted the question of why PPO resides in this particular cellular compartment, as physical separation from its substrates can also be achieved in the cytosol, and therefore if there is indeed a role for PPO independent of tissue damage (Yoruk and Marshall, 2003; Thipyapong et al., 2004b; Boeckx et al., 2015b). Many phenolics are antioxidants (Pietta, 2000; Parveen et al., 2010), acting as radical scavengers (Neill and Gould, 2003; Agati et al., 2007) and photochemical energy dissipaters (Grace and Logan, 2000; Neill and Gould, 2003) which could contribute to the protection of the chloroplast against stress.

In this study, phenolic contents of purified chloroplasts were analyzed from 12 week old mutant and wild-type plants in a two phase cross over design. The chloroplastic metabolites, coumaroyl hexoside, coumaroyl malate, and phaselic acid (caffeoyl malate), were detected in both phases of the cross over design (Figure 6).

In mutant plants, none of the three chloroplastic metabolites differed significantly (Figure 6). However, in wild-type purified chloroplasts both coumaroyl hexoside and phaselic acid accumulated as compared with the mutant, although during different phases of the experimental design (Figure 6), and this accumulation appeared to be light dependent. Coumaroyl hexoside was detected at elevated levels in the dark in both phases of the cross over design; though it was significantly higher in the first phase only (p < 0.05; Figure 6A). No differences were observed in accumulation of coumaroyl malate (Figure 6B), while an increase in phaselic acid, detected in the second (light) phase (p < 0.01; Figure 6C), strongly indicated that this was an effect of PPO activity in illuminated, wild-type chloroplasts, consistent with the findings of Halliwell (1975). A proposed pathway for the conversion of coumaroyl hexoside to phaselic acid is presented in Figure 7. In this the light independent conversion of coumaroyl hexoside to coumaroyl malate is catalyzed by hydroxycinnamoylglucose:malate hydroxycinnamoyltransferase (HMT; Lehfeldt et al., 2000) and the light requiring conversion of coumaroyl malate to phaselic acid is catalyzed by PPO. It is recognized that due to the complexity of interconversion of secondary metabolites other enzymes may also be involved in vivo.

These chloroplastic coumaroyl conjugates can act as substrates for red clover PPO with monophenolase activity. While coumaroyl hexoside was barely detectable in total leaf extracts (data not shown); it was clearly detectable in many chloroplastic extracts, suggesting a localized accumulation perhaps under specific environmental conditions (Figure 6C).

Here, the red clover plants were not visibly stressed, yet removal of leaves after the first of the two phase cross over design could affect plant metabolism during the second phase (via a systemic damage response). It may also remove shading from some of the leaf population. Red clover PPO activity has been shown to increase as young leaves expand and therefore become exposed to higher light intensities (Webb et al., 2013), which might further implicate a role for PPO in protecting the developing photosynthetic apparatus.

The findings reported here are consistent with hydroxylation of the monophenolic coumaroyl conjugate (malate) to a caffeoyl conjugate (malate) in stressed plants in the light. Thus, there may have been conversion of coumaroyl malate to caffeic malate (phaselic acid) by PPO monophenolase in these stressed, illuminated chloroplasts of wild-type plants. Coumaroyl hexoside may be an intermediate metabolite.

In support of this hypothesis is a recently published report on the hypothetical role for walnut (Juglans regia) PPO in secondary metabolism (Araji et al., 2014). However, they observed an altered leaf metabolite profile in PPO-silenced walnut compared to the wild-type (Araji et al., 2014), by contrast, no differences between the wild-type and the low leaf PPO mutant population of red clover in terms of phenolic profile were observed in this study (Figure 2).

In conclusion, metabolic profiling has enabled the identification of compounds with the potential to be chloroplast localized PPO substrates. While the hydroxycinnamic acid conjugates require further functional testing, their presence in red clover chloroplasts indicates that in undamaged green tissues, secondary metabolism in chloroplasts could be a neglected component of abiotic stress tolerance. The identification of coumaroyl hexoside and coumaroyl malate in chloroplasts and its apparent conversion to caffeic malate (phaselic acid) in chloroplasts of stressed leaves in the light further highlights the need to continue our efforts in understanding the complexity of secondary metabolism and compartmentation in relation to a role for PPO in undamaged tissues. If PPO activity has a role in protection of the photosynthetic apparatus against damage under abiotic stress conditions, this would make PPO activity a potential breeding target for improved performance of food and bioenergy crops in the future.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.