• The differential accumulation of various polyphenols, particularly of flavonoids and hydroxycinnamates, was studied in leaves of Ligustrum vulgare exposed to increasing sunlight under well watered or drought-stress conditions.
• Light- and drought-induced changes in leaf polyphenol concentrations were normalized to the CO2 assimilation rate. The functional roles of flavonoids and hydroxycinnamates were analysed through tissue localization using multispectral fluorescence microimaging, and through efficiencies to scavenge superoxide radicals (O2−) and to screen UV wavelengths.
• Clear effects of light and water treatments on leaf polyphenol concentrations were not observed, as the CO2 assimilation rate varied according to sunlight and water availability. However, biosynthesis of quercetin 3-O-rutinoside, luteolin 7-O-glucoside and echinacoside, which were efficient O2− scavengers, increased sharply in response to solar radiation. By contrast, carbon for the synthesis of p-coumaric acid and monohydroxyflavones, efficient UV screeners but poor O2− scavengers, did not vary depending on light treatments. Flavonoids accumulated in both the adaxial epidermis and the palisade tissue because of sunlight irradiance, whereas echinacoside occurred largely in abaxial tissues.
• We hypothesize that flavonoids may serve antioxidant functions in response to excess light and drought stress, and that a coordinate control system between hydroxycinnamate and flavonoid pathways operated in L. vulgare exposed to excess light.
In the present experiment, conducted on Ligustrum vulgare leaves exposed to increasing solar radiation under well watered or drought-stressed conditions, we (i) identified and quantified the whole polyphenol spectrum; (ii) related light- and drought-induced changes in polyphenol concentration to corresponding changes of net CO2 assimilation; (iii) determined whether a relationship exists between the differential accumulation of hydroxycinnamates and flavonoids and their UV-absorbing or in vitro antioxidant properties (scavenger activity against superoxide radicals); and (iv) analysed the tissue-specific distribution of flavonoids and hydroxycinnamates. Ligustrum vulgare, along with other members of the Oleaceae family distributed in the Mediterranean basin, may represent an interesting species for elucidating the role of individual polyphenol classes in the response mechanisms to various environmental stimuli, in view of a complex polyphenol and particularly flavonoid composition (Pieroni et al., 2000; Romani et al., 2000; Tattini et al., 2000).
Materials and Methods
Plant material and growing conditions
One-yr-old container-grown Ligustrum vulgare L. plants were headed back to four 15 cm shoots at the end of May at Pisa, Italy (43°43′ N, 10°23′ E). Plants were grown outdoors under 20–25% sunlight radiation and supplied with 1/3-strength Hoagland's solution over a 3 wk period. At that time plants had produced three nodes, and the youngest leaves, the area of which accounted for approx. 25% of leaf lamina sizes at full development, were labelled for all subsequent measurements.
Plants were then placed in 100 m2 boxes constructed with black polyethylene frames to receive 6% (shade), 35% (mid-sun) or 100% (sun) sunlight radiation. Sunlight shading was estimated over the 300–1100 nm waveband, using a LI-1800 spectroradiometer (LI-COR Inc., Lincoln, NE, USA) equipped with a remote cosine receptor. The daily dose of UV-B irradiance was measured by an SUV100 scanning spectroradiometer (Biospherical Instruments, San Diego, CA, SA) on a total of 20 d, both clear and cloudy, over an 8 wk experimental period. Plants at the full-sun site received a daily dose of 11.4 and 0.98 MJ m−2 and 18.9 KJ m−2 in the PAR (photosynthetic active radiation over 400–700 nm), UV-A and UV-B wavebands, respectively. These data were in good agreement with daily courses of irradiance obtained from the ELDONET dosimeter located at the Pisa station (Marangoni et al., 2000). Mean daily doses of 0.29 and 0.05 MJ m−2 in the UV-A waveband, and of 6.0 and 0.8 KJ m−2 in the UV-B waveband, were recorded at the mid-sun and shade sites, respectively.
Plants at the three sites were supplied with either 100% (well watered) or 40% (drought-stressed) of the daily evapotranspiration demand (evapotranspiration was determined by measuring daily weight loss of five replicate pots, which were irrigated until complete leaching of the substrate), over an 8 wk experimental period. The drought-stress treatment was achieved by progressive reductions of daily water supply during a 5 d period, when the experiment started. Thus leaves labelled for measurements had been exposed to contrasting sunlight irradiance for 5 d at the beginning of the experiment.
Identification and quantification of soluble polyphenols
Extraction and purification procedures, both HPLC-DAD and HPLC-MS analyses, were performed similarly to the protocol previously reported for Ligustrum sinensis leaves (Romani et al., 2000). In brief, 30 mg freeze-dried tissue was extracted with 3 × 8 ml EtOH/H2O solution (75/25; v/v) adjusted to pH 2.5 with formic acid. The supernatant was then partitioned with 3 × 8 ml n-hexane, reduced to dryness under vacuum (Rotavapor 144R, Büchi, Flawil, Switzerland) at 25°C, and the ethanol fraction diluted with 1 ml H2O/MeOH/CH3CN (20/60/20; v/v/v). Aliquots 10–25 µl were injected in an HP1100 liquid chromatograph equipped with a diode array detector (DAD), and managed by an HP workstation (all from Hewlett Packard, Palo Alto, CA, USA). The column was a Merck LiChrosorb RP18 (Merck, Darmstadt, Germany), 4.6 × 250 mm ID, kept at 26 ± 1°C, and equipped with a 10 × 4 mm LiChrosorb RP18 precolumn. The eluent was H2O (pH 3.2 by H3PO4)/CH3CN, and polyphenols separated by using the seven-step linear gradient solvent system with a flow rate of 1 ml min−1 during a 106 min run, previously developed by Romani et al. (1996). Analysis of retention times, and both spectroscopic and spectrometric data, fully identified individual polyphenols. HPLC-MS analysis was performed by interfacing the HPLC-DAD system with a Hewlett Packard 1100 MSD API-electrospray operating in the negative ion mode as reported by Romani et al. (2002). Echinacoside, a caffeic glycoside ester previously detected in other Oleaceae spp. (Andary et al., 1992), was isolated by semipreparative HPLC (Romani et al., 2002). Quantification of individual compounds was carried out using five-point calibration curves (r2 > 0.99) operating in the range 0–40 g. Secoiridoids and tyrosol derivatives were calibrated at 280 nm, p-coumaric acid and echinacoside at 315 and 330 nm, respectively, and calibration of individual flavonoid glycosides was carried out at 350 nm.
Gas-exchange measurements and calculation of CO2-based polyphenol accumulation
Diurnal variation in the net CO2 assimilation rate was measured using a LI-COR 6400 IRGA analyser operating at ambient CO2 of 34 ± 0.5 Pa. Measurements were conducted at 3 h intervals from 06 : 00 to 21 : 00 h over 2 consecutive days. Daily assimilated CO2 was then calculated by integrating the diurnal curve as described by Valladares & Pearcy (1997) using Sigma Plot (2000) ver. 8 software (SPPS Inc., Chicago, IL, USA). The daily course of net CO2 assimilation rates was determined at weekly intervals (on a total of 16 d over the experimental period), and total assimilated CO2 was then estimated by plotting the mean daily net CO2 assimilation vs time (wk) using the integration procedure reported above.
The CO2-based polyphenol (poly) accumulation () was then calculated by normalizing the increase of polyphenol concentration on a leaf-area basis (mol poly m−2) to the assimilated CO2 during the experimental period (t0 to t1), using the following equations:
mol poly m−2 = mol poly g−1 d. wt × g d. wt m−2(Eqn 1)
where A = mol poly m−2 at t1, B = mol poly m−2 at t0 and C is the CO2 assimilated over the whole experimental period t0 to t1. Leaf mass per area (LMA, g d. wt m−2) was measured as previously reported (Tattini et al., 2000). The polyphenol concentration of the youngest leaves that had been exposed to shade, mid-sun and sun conditions for 5 d averaged 18.4, 23.2 and 27.8 µmol g−1 d. wt, respectively, at t0. At that time, LMA of shade, mid-sun and sun leaves averaged 28.7, 30.2 and 31.5 g d. wt m−2, respectively.
UV-spectral and in vitro antioxidant properties of individual hydroxycinnamates and flavonoids
The UV-absorption spectra of 50 m ethanol solutions of hydroxycinnamates and flavonoids were recorded using a Jasco 560 V UV-Vis spectrophotometer (Jasco, Tokyo, Japan). The in vitro antioxidant activities of individual metabolites were estimated in terms of their scavenger activities against the superoxide (O2−) radical, using electronic paramagnetic resonance (EPR) measurements. Superoxide radicals were generated by adding 100 l 0.3 m H2O2 and 70 l 0.4 mm KOH (blank) or an equal volume of metabolite solution (in KOH 0.4 mm, in the concentration range 5–500 m) to 130 l 0.4 m KOH. The reaction was started with the addition of 100 l acetone and stopped by freezing the quartz test tubes in liquid nitrogen after 30 s, as reported previously (Sichel et al., 1991, Baratto et al., 2003). The EPR spectra were recorded on a Bruker 200D SRC X-band spectrometer, under the following operating conditions, at 120K: microwave frequency 9.565 GHz, centre field 335 mT, scan width 50 mT.
The scavenger activities, on a percentage basis, of individual hydroxycinnamates and flavonoid glycosides were then calculated by changes of EPR signal intensity using the equation:
R = [(H0 − Hx)/H0] × 100(Eqn 3)
where R is the normalized scavenger ratio, and H0 and Hx are the EPR signal intensities of blank or sample solutions, respectively. The half-inhibition concentration (IC50) of individual hydroxycinnamates and flavonoids against O2− was then calculated by plotting the normalized scavenger ratios vs metabolite concentration (Baratto et al., 2003). Raw data were fitted (r2 > 0.98 for all the tested metabolites) by second-order linear regression equations (Sigma Plot 2000 ver. 8). Superoxide radicals were also generated using the xanthine/xanthine oxidase (X/XO) system, and scavenger activities of hydroxycinnamates and flavonoids were determined spectrophotometrically at 560 nm using the NBT assay (Baratto et al., 2003). The order of scavenger activities of individual metabolites against O2− did not differ when spectrophotometric data were compared with EPR measurements.
Fluorescence microspectroscopy and multispectral fluorescence microimaging
Transverse sections (50 µm thick) of fresh leaf tissue were cut with a vibratory microtome (Vibratome 1000 Plus, Vibratome, St Louis, MO, USA) at a point approximately one-third of the lamina length from the tip, and incubated in 0.1% (w/v) 2-amino ethyl diphenyl boric acid (Naturstoff reagent, NR) in phosphate buffer (pH 6.8) with addition of 1% NaCl (w/v), as reported by Hutzler et al. (1998). Staining time was 5 min, and excess NR was removed by washing samples with phosphate buffer for 2 min. Fluorescence microspectroscopy and multispectral fluorescence microimaging were performed using a standard inverted epifluorescence microscope (Diaphot, Nikon, Japan), coupled to both a charge-coupled device (CCD) camera and a multichannel spectral analyser, as described by Agati et al. (2002). Fluorescence spectra of adaxial and abaxial epidermal, and both palisade and spongy parenchymal tissues, were recorded at UV (λexc = 365 nm) and blue-light (λexc = 470 nm) excitation wavelengths, as detailed in Appendix 1. Then fluorescence images at 580 nm (F580) were sequentially recorded of UV- (λexc = 365 nm, ) and blue-light – (λexc = 470 nm, ) excited cross-sections, by means of a motorized filter wheel coupled to the CCD detector. The tissue-specific distribution of flavonoids and hydroxycinnamates was finally determined by analysing F580 profiles, from adaxial to abaxial epidermal tissues, as described in Appendix 1.
Experimental design and statistics
The experiment was a completely random design. The diel variation of net CO2 assimilation rate was determined on three leaves per plant, on five plants per treatment, at weekly intervals. Leaf mass per area (LMA) was measured on five leaves per treatment for a total of 10–15 leaf discs at the beginning (t0) and end (t1) of the experiment. Polyphenol concentration was measured for three replicate samples, each comprising five to six undeveloped leaves, at t0. At t1, leaves for gas-exchange measurements were sampled (five replicates of three leaves each) to determine the polyphenol content. CO2-based polyphenol accumulation was then calculated, normalizing changes in polyphenol concentration to total assimilated CO2 of the youngest undeveloped leaves, exposed to 8 wk of light and water treatments. In vitro scavenger activities of individual hydroxycinnamates and flavonoids against superoxide radicals were determined in triplicate. Multispectral fluorescence microscopy was conducted on two transverse sections per leaf, on three leaves per treatment, at the end of the experimental period. A two-way anova (factors light and water) was performed to estimate significant treatment effects on both total and individual polyphenol concentrations (Statgraphic Plus ver. 5.1, Manugistic Inc., Rockville, MD, USA).
The polyphenol concentration of L. vulgare was markedly affected by light and water treatments, with significant differences between shade leaves (6% sunlight) and mid-sun leaves (35% sunlight), and very little variation between mid-sun and sun (100% sunlight) leaves (Fig. 1a). Drought stress significantly decreased the polyphenol concentration of leaves developed at 35 or 100% solar radiation, whereas it did not change that of leaves developed at the shade site (Fig. 1a). Light and water treatments also dramatically affected the CO2 assimilation rate of L. vulgare leaves (Fig. 1b). In detail, the daily carbon gain in drought-stressed leaves accounted for 71, 49 and 60% of that in well watered leaves at the shade, mid-sun and sun sites, respectively. Diel net CO2 assimilation rate also varied dramatically because of sunlight irradiance, as the daily carbon gain of well watered shade and full-sun leaves accounted for only 23 and 80%, respectively, of diel CO2 assimilation of mid-sun leaves (Fig. 1b). Furthermore, LMA increased steeply, passing from shade (44.0 g d. wt m−2) to mid-sun (75.1 g d. wt m−2) and full-sun plants (99.2 g d. wt m−2), whereas it was unaffected by water treatments (data not shown). As a consequence, polyphenol concentration differed markedly from CO2-based polyphenol () accumulation of L. vulgare leaves developed at contrasting light and water availabilities (Fig. 1a,c). For example, of shade and mid-sun leaves did not differ appreciably, but was markedly smaller than of leaves exposed to 100% sunlight. Remarkably, the in drought-stressed leaves was substantially greater than in well watered leaves (particularly evident on leaves exposed to 6% sunlight), although drought stress had significantly decreased the leaf polyphenol concentration of L. vulgare (Fig. 1a,c).
HPLC-DAD profiles reported in Fig. 2 show the complex polyphenol composition of L. vulgare leaves, which included secoiridoids and tyrosol derivatives, and both hydroxycinnamates and flavonoid glycosides. Oleuropein, ligustaloside A and ligustaloside B, and ligstroside constituted the relevant class of secoiridoids (Fig. 2b), which appeared more complex than that previously detected in other members of the Oleaceae (Pinelli et al., 2000; Tattini et al., 2000). The flavonoid composition of L. vulgare leaves also appeared of particular interest (Fig. 2a) as it comprised the flavonol quercetin 3-O-rutinoside (que 3-O-rut); two luteolin glucosides, namely the dihydroxyflavone luteolin 7-O-glucoside (lut 7-O-glc) and the monohydroxyflavone luteolin 4′-O-glucoside (lut 4′-O-glc); and both apigenin 7-O-glucoside (api 7-O-glc) and apigenin 7-O-rutinoside (api 7-O-rut). Hydroxycinnamates were both p-coumaric acid and echinacoside, a rare caffeic glycoside ester, which was fully identified by HPLC-MS analysis. Appreciable amounts of hydroxytyrosol and hydroxytyrosol glucoside were also detected in L. vulgare leaves, as previously reported to occur in analogous tissues of Olea europaea (Pinelli et al., 2000).
Possible roles served by various polyphenols in the response mechanisms of L. vulgare leaves to excess light and drought stress were analysed by changes in both concentration and CO2-based accumulation of different metabolites in response to increasing sunlight and drought stress (Figs 3, 4). The secoiridoid concentration increased sharply passing from shade to mid-sun or sun leaves, but the relative CO2-based accumulation rates showed less evident variations (Fig. 3a,c). Similarly, the drought-induced decrease of secoiridoid concentration was mainly caused by markedly different leaf CO2 assimilation rates between drought-stressed and well watered L. vulgare plants. Light and water treatments did not alter the secoiridoid composition of L. vulgare leaves (data not reported). The concentration of tyrosol derivatives did not vary in leaves exposed to increasing sunlight, whereas it was significantly decreased in drought-stressed leaves (Fig. 3b). Remarkably, the contribution (percentage basis) of secoiridoids + tyrosol derivatives to total polyphenol content substantially decreased in response to increasing sunlight radiation (from 59% in shade to 36% in sun leaves) and drought stress (from 51% in well watered to 36% in drought-stressed leaves).
Hydroxycinnamates and, above all, flavonoid levels in L. vulgare leaves were most strongly affected by light and drought stress (Fig. 4). However, a decreased amount of assimilated carbon was devoted to the synthesis of p-coumaric in response to increasing sunlight, although the leaf concentration of p-coumaric increased because of solar radiation (Fig. 4a,g). By contrast, carbon usage for echinacoside synthesis was markedly increased by excess light and drought stress in L. vulgare leaves (Fig. 4h). As a consequence, the light-induced decrease of the ratio of hydroxycinnamate to total polyphenol (from 0.42 in shade to 0.36 and 0.31 in mid-sun and sun leaves, respectively) was exclusively caused by light-induced changes of p-coumaric content (Fig. 4a). On the whole, the composition and concentration of flavonoids were mainly affected by light treatments in L. vulgare leaves (Fig. 4c–f), as the concentrations of que 3-O-rut and lut 7-O-glc (Fig. 4c,d), but not those of monohydroxyflavones, i.e. lut 4′-O-glc and apigenin glycosides (Fig. 4e,f), increased because of solar radiation. Different accumulations of api 7-O-rut and api 7-O-glc in response to light and water treatments were not detected (data not reported). Ligustrum vulgare leaves exposed to drought stress devoted a substantially greater amount of carbon for the synthesis of que 3-O-rut and lut 7-O-glc than did well watered leaves (particularly evident in leaves grown at 6% sunlight; Fig. 4i,j). The preferential carbon usage for the synthesis of echinacoside, que 3-O-rut and lut 7-O-glc at the expense of p-coumaric acid and monohydroxyflavone synthesis, in response to light and drought stress, was strongly correlated to the in vitro scavenger activities of individual metabolites against O2− (Fig. 5). Only those metabolites with a catechol group in the benzene ring of the hydroxycinnamate skeleton, as in echinacoside (Fig. 5b), or in the B ring of the flavonoid skeleton, as in que 3-O-rut and lut 7-O-glc (Fig. 5c,d), were efficient scavengers of O2−. On the other hand, the UV-absorbing features of highly efficient O2− scavengers (IC50 for O2− < 100 m; Fig. 5b–d) did not differ from those of monohydroxyflavones (Fig. 5e,f) and p-coumaric acid (Fig. 5a), which were poor quenchers of superoxide radicals (IC50 for O2− > 500 m).
The distribution of flavonoids and hydroxycinnamates in epidermal and mesophyll tissues of L. vulgare leaves exposed to increasing solar radiation has been reported in Fig. 6 by analysing the fluorescence spectra of UV-excited cross-sections (Fig. 6a–c) and the F580 profiles under both UV and blue-light excitations (Fig. 6d–f). The shape of normalized fluorescence spectra (Appendix 1; Cerovic et al., 1999; Agati et al., 2002) of L. vulgare leaves exposed to increasing sunlight depended almost exclusively on the tissue concentration of echinacoside, que 3-O-rut and lut 7-O-glc (Appendix 1; Agati et al., 2002).
On the whole, the contribution of flavonoids to the yellow-red fluorescence of L. vulgare leaves increased because of sunlight irradiance, whereas the contribution of blue fluorescing compounds (both wall-bound and soluble hydroxycinnamates) to the shape of tissue fluorescence spectra was noticeably greater in shade and mid-sun than in sun leaves. Nevertheless, flavonoids almost exclusively contributed to the fluorescence signatures of the adaxial epidermal layer of both mid-sun and sun leaves, as revealed by the peak of maximum emission at 570–575 nm (Fig. 6a,b) and the high values (Fig. 6d,e). Flavonoids scarcely occurred in the palisade tissue of mid-sun leaves (peak of maximum emission at 552 nm), whereas they accumulated (peak at 572 nm) in analogous tissues of sun leaves. Hydroxycinnamates, namely echinacoside (Appendix 1), appreciably occurred in the adaxial palisade parenchyma of mid-sun leaves, and were scarcely distributed in the majority of palisade parenchymal cells of sun leaves (Fig. 6d,e). On the whole, a nearly exclusive accumulation of echinacoside was detected in the spongy parenchyma of mid-sun and sun leaves, as revealed by (i) the emission peaks in the green-yellow waveband (at 530 and 540 nm for mid-sun and sun leaves, respectively); (ii) the relevant contribution of blue-fluorescing compounds (emitting in the 450–480 nm waveband) to the tissue fluorescence spectra (Fig. 6a,b); and (iii) the greatest nonflavonoid emission coupled with negligible values (Fig. 6d,e). The fluorescence signatures of abaxial epidermis of flatly angled (leaf angle 10 ± 3°) mid-sun leaves (peaks at 480 and 530 nm; Fig. 6b) mostly depended on the occurrence of blue-green fluorescing compounds, whereas flavonoids largely accumulated in analogous tissues of steeply angled (leaf angle 58 ± 6°) sun leaves (peak of maximum emission at 575 nm; Fig. 6a). Finally, blue-green fluorescing compounds made the greatest contribution to the fluorescence signatures of shade leaves, whereas a scarce accumulation of flavonoids was limited to the adaxial epidermal layer (Fig. 6c,f).
Data from the experiments described here, which result from measurements conducted at different scale levels, address some open questions concerning the roles played by hydroxycinnamates and flavonoids in the response mechanisms of plants exposed to excess light and drought stress.
The first contribution made by the present work is essentially methodological. We demonstrate that light- or drought-induced changes in the synthesis of polyphenols cannot be directly estimated by the mere analysis of leaf polyphenol concentration, as both stressful conditions may concomitantly affect leaf CO2 assimilation rate (Hampton, 1992; Agrell et al., 2000; Jifon & Syvertsen, 2003) and leaf construction (LMA, Evans & Poorter, 2001). We present evidence that the strikingly different polyphenol concentrations of L. vulgare leaves exposed to 6 or 35% solar radiation were a result of light-induced variations on total assimilated carbon, rather than on the share of carbon devoted for their synthesis (Fig. 1a,c). Furthermore, the polyphenol content of leaves suffering from excess light at the full-sun site (L. vulgare occurs in partially shaded areas of the Mediterranean basin; Brosse, 1979) did not differ from that of leaves grown under 35% solar irradiance, but sun leaves devoted a substantially greater amount of assimilated carbon to the synthesis of polyphenols than did mid-sun leaves (Fig. 1a,c). The CO2-based polyphenol accumulation proposed in this work (Eqn 1) is still a rough estimation of the carbon actually available for the synthesis of secondary metabolites, as the share of daytime assimilated carbon used in the dark respiration, which may vary in leaves exposed to both contrasting light and water regimes (Amthor, 1995; Green & Lange, 1995), has not been included in our equations (Eqns 1, 2). Nevertheless, it is a major advantage of that it is independent of the stress severity (intensity and duration) and of interspecific tolerances to environmental injury (Olsson et al., 1998; Nogués & Baker, 2000; Alexieva et al., 2001; Jifon & Syvertsen, 2003). As a consequence, may be a valuable tool for revisiting most of the conflicting data from the literature (Stephanou & Manetas, 1997; Pääkkönen et al., 1998; Estiarte et al., 1999; Nogués & Baker, 2000).
Second, we highlight that only flavonoids (que 3-O-rut and lut 7-O-glc) and hydroxycinnamates (echinacoside) which were efficient scavengers (in vitro) of superoxide radicals served a role in the response mechanisms of L. vulgare leaves exposed to drought stress and increasing sunlight radiation, confirming most recent reports on this matter (Ryan et al., 1998; Agati et al., 2002; Jordan, 2002; Hofmann et al., 2003). Scavenger activities of flavonoids and hydroxycinnamates against superoxide radicals (Fig. 5) that exclusively resulted from the relative ability to quench O2− using the EPR technique (flavonoids may additionally inhibit the activity of xanthine oxidase in NBT assay; Chang et al., 1993; Baratto et al., 2003), were in the order: flavonol (que 3-O-rut) > dihydroxyflavone (lut 7-O-glc) > echinacoside >> monohydroxyflavones (lut 4′-O-glc and api 7-O-gly). Our findings agree with well known relationships between chemical structure and antioxidant activity of polyphenols in vitro (Fig. 5; Hu et al., 1995; Rice-Evans et al., 1997). Remarkably, the accumulation of ‘antioxidant’ flavonoids occurred not only in L. vulgare leaves exposed to 35 and 100% solar radiation, but also in shade leaves suffering from drought stress (Fig. 4c,d). These data, taken together with the similar UV-absorbing features of all the flavonoids detected in L. vulgare leaves, demonstrate that flavonoids may serve a key role as free radical scavengers in the response mechanisms to excess light stress (Ryan et al., 1998; Gould et al., 2000; Tattini et al., 2000). It should also be noted that the exclusive accumulation of que 3-O-rut and lut 7-O-glc in the general biochemical strategy of L. vulgare in response to excess sunlight irradiance may additionally relate to the greater abilities of flavonol and dihydroxyflavones than monohydroxyflavones in dissipating (through tautomerization) potentially harmful UV-B radiation (Smith & Markham, 1998).
The light-induced decrease of hydroxycinnamate : flavonoid ratio observed in L. vulgare leaves (Fig. 4) confirms previous suggestions of a minor role served by the whole class of hydroxycinnamates in the response mechanisms to excess UV radiation (Olsson et al., 1999; Burchard et al., 2000; Hoffmann et al., 2003). Nevertheless, the dramatic shift of carbon usage for the synthesis of echinacoside at the expense of p-coumaric acid in L. vulgare exposed to both excess light and drought stress, agree with a key protective function of caffeic esters in leaves exposed to different stressful agents (Grace et al., 1998; Schoch et al., 2001; Nair et al., 2002). Finally, the role served by secoiridoids and tyrosol derivatives in the response mechanisms of L. vulgare to excess light and drought stress appears of minor significance, as their contribution to the total polyphenol pool decreased steeply because of sunlight and drought stress (Figs 3, 4). Such a suggestion appears to be further corroborated by the low efficiency of such phenolics in absorbing short solar wavelengths (relative absorption maxima between 240 and 280 nm). However, secoiridoids, particularly oleuropein, which are highly efficient deterrents against predators (Konno et al., 1999), may increase the leaf lifespan of L. vulgare plants exposed to excess light and drought stresses (Aerts, 1995).
The final contribution of this work relates to the development of a multispectral fluorescence microimaging technique (Fig. 6) to analyse the localization–functional relationships of both hydroxycinnamates and flavonoids in leaf tissues exposed to increasing doses of solar radiation (Olsson et al., 1998; Olsson et al., 1999; Gould et al., 2000). Our fluorescence image approach represents a valuable complement to both ‘destructive’ (Ålenius et al., 1995; Olsson et al., 1999) and chlorophyll fluorescence-based techniques (Bilger et al., 2001; Cerovic et al., 2002), which limit the analysis of phenylpropanoid distribution to the level of epidermal tissues. Here it is conclusively shown that mesophyll flavonoids, not only epidermal flavonoids, are closely related to sunlight exposure of L. vulgare leaves (Fig. 6d–f). The accumulation of flavonoids () in the palisade tissue of leaves exposed to 6 and 35% sunlight accounted for 10 and 42% of of leaves grown at 100% solar radiation. The light-induced increase of mesophyll flavonoids should have resulted from an enhanced oxidative load (Yamasaki et al., 1997; Jordan, 2002), as previously reported to occur in leaves exposed to both UV and solar radiation (Knogge & Weissenböck, 1986; Liu et al., 1995; Tattini et al., 2000; Agati et al., 2002). As a consequence, the large distribution of que 3-O-rut and lut 7-O-glc in the palisade cells of L. vulgare leaves exposed to 100% sunlight correlates with their in vitro scavenger activities. However, flavonoid glycosides have been shown to be located predominantly in the cell vacuole, and are unlikely to encounter superoxide radicals in such a compartment (Olsson et al., 1998; Neill & Gould, 2003). Nevertheless, the ‘antioxidant’ properties of que 3-O-rut and lut 7-O-glc might be of increasing significance under conditions of severe light stress, when H2O2 (formed from O2− by the action of superoxide dismutase) leaks from cellular organelles and freely diffuses across the tonoplast to enter the vacuole (Yamasaki et al., 1997; Olsson et al., 1998; Neill & Gould, 2003). Interestingly, our results do not confirm previous suggestions that mesophyll flavonoids are mainly constitutive and undergo slight modifications under increasing UV radiation (Reuber et al., 1996; Burchard et al., 2000). We suggest that large variations of both constitutive and light-induced changes of leaf morpho-anatomical features of investigated species (Lovelock et al., 1992; Mittler, 2002), coupled with strikingly different light treatments (Caldwell et al., 1995; Tattini et al., 2000), were likely to be responsible for such contrasting results.
The tissue-specific distribution of nonflavonoids ( nonflavonoids, Fig. 6d–f) was consistent with both the concentration and the percentage contribution of echinacoside to the pool of fluorescing compounds in shade (84%), mid-sun (51%) and sun (45%) leaves, and confirms previous findings of a preferential distribution of hydroxycinnamates in abaxial tissues of UV-B treated leaves (Ålenius et al., 1995; Olsson et al., 1999). However, the nearly exclusive accumulation of echinacoside in the inner spongy cells of sun leaves (Fig. 6d) appeared poorly related to its efficiency as both O2− scavenger and UV absorber (Fig. 5b). It may be argued that the flavonoid pathway has been favoured with respect to the hydroxycinnamate pathway (Li et al., 1993; Bate et al., 1994) in epidermal and palisade cells of L. vulgare leaves developed at the sun site. It has recently been suggested that synthesis of caffeic derivatives, starting from p-coumarate, originates from substrates that give priority to the synthesis of flavonoids, in tissues exposed to severe stress conditions (e.g. wounding; Schoch et al., 2001). The exclusive occurrences of flavonoids in the abaxial epidermal layer of steeply angled sun leaves (Fig. 6a,d) and of blue-green fluorescing hydroxycinnamates in the analogous tissue of flatly angled mid-sun leaves (Fig. 6b,e) strongly supports the idea that a coordinated control system between hydroxycinnamate and flavonoid pathways (Christensen et al., 1998; Blount et al., 2000; Ruegger & Chapple, 2001) operates in L. vulgare exposed to excess light.
We wish to thank Maria Laura Traversi for her valuable help in gas-exchange measurements and cross-sectioning.
Equipment and theoretical background for fluorescence microspectroscopy and multispectral fluorescence microimaging
Fluorescence microspectroscopy Fluorescence measurements were performed using an inverted epifluorescence microscope (Diaphot, Nikon, Japan) equipped with a high-pressure mercury lamp (HBO 100 W, OSRAM, the Netherlands) as light source. The excitation wavelengths were selected using 10 nm bandwidth interference filters 365FS10-25 and 470FS10-50 (Andover Corporation, Salem, NH, USA), and dichroic mirrors ND400 and ND510 (Nikon) for the 365 and 470 nm wavelengths, respectively. Fluorescence spectra were measured by a CCD multichannel spectral analyser (PMA 11-C5966, Hamamatsu, Photonics Italia, Arese, Italy) connected to the microscope through an optical fibre bundle, using a ×10 Plan Fluor objective which integrated the fluorescence intensity on a 0.0078 mm2 area. Residual excitation light was removed by GG400 and GG495 long-pass filters (Schott Glas, Mainz, Germany) for the 365 and 470 nm excitation wavelengths, respectively. Fluorescence spectra (over the 300–800 nm waveband), with an excellent signal-to-noise ratio, were integrated over 1 s. Several fluorescence measurements were recorded for the same tissue because of the very short excitation time, which preserved tissues from photo damage. In detail, a total of 20 fluorescence spectra of each individual tissue layer – adaxial and abaxial epidermal, and both palisade and spongy parenchymal tissues – were recorded sequentially. Finally, fluorescence spectra were averaged and corrected for the transmission spectra of optics and filters, and reported in smoothed form (Fig. 6a–c).
Fluorescence features of L. vulgare tissues were analysed using normalized fluorescence spectra, as changes in fluorescence intensity caused by variations in the tissue anatomy (McClendon & Fukshansky, 1990) were removed following normalization (Agati et al., 2002). It has been shown previously that the relative contributions of hydroxycinnamates and flavonoids, both wall-bound and soluble, determine the shape of normalized fluorescence spectra under UV excitation (Buschmann & Lichtenthaler, 1998; Agati et al., 2002). In this experiment the contribution of wall-bound phenolics to light-induced changes of leaf fluorescence signatures was negligible, as (i) we did not observe appreciable changes in the content of wall-bound phenolics of leaves exposed to increasing sunlight radiation (data not shown), as previously reported to occur in another member of the Oleaceae (Agati et al., 2002); and (ii) excitation of wall-bound phenolics was negligible using the 365 nm excitation wavelengths available in our microscope.
Soluble polyphenols actually contributing to the fluorescence signatures of UV-excited and NR-stained cross-sections of L. vulgare leaves were que 3-O-rut and lut 7-O-glc (maximum emission over the yellow-red (580–605 nm) waveband; Agati et al., 2002) and echinacoside (maximum emission in the green waveband at 530 nm, similar to the emission peak of verbascoside; Agati et al., 2002). In fact, p-coumaric acid, as other mono-hydroxycinnamates, did not fluoresce at λexc = 365 nm (as may be also argued from the UV-absorption spectrum of p-coumaric acid; Fig. 5a). Furthermore, fluorescence quantum yields of apigenin derivatives (both api 7-O-glc and api 7-O-rut), lacking the ortho-dihydroxy structure in the B-ring of the flavonoid skeleton, did not exceed 8% of fluorescence quantum yields of lut 7-O-glc (Agati et al., 2002). We also observed that the fluorescence characteristics of the monohydroxyflavone lut 4′-O-glc did not differ from those of apigenin glycosides (data not shown). As a consequence mono-hydroxyflavones, concentrations of which did not vary in response to light treatments (Fig. 4), did not contribute to the fluorescence signatures of L. vulgare leaves exposed to increasing sunlight. Finally, secoiridoids and tyrosol derivatives did not make an appreciable contribution to fluorescence signatures of L. vulgare leaves because of the negligible quantum yields of both polyphenol classes (Agati et al., 2002).
Multispectral fluorescence microimaging The tissue specific distribution of fluorescing metabolites in L. vulgare leaves was analysed using the following methodology. Fluorescence images at 580 nm (F580) were sequentially recorded for UV- (λexc = 365 nm, ) and blue-light-excited (λexc = 470 nm, ) cross-sections treated with NR. Images were collected by a CCD camera (Chroma CX260, DTA, Italy) equipped with a Kodak KAF261E (512 × 512 pixels) detector, digitized with 14-bit (16 384 grey levels) dynamics and saved as tagged image files (TIF). Image processing included background subtraction, flat-field correction for spatial nonuniformity of the excitation beam, and sharpen filtering (Murphy, 2001). A representative F580 image, excited with blue light at 470 nm, of L. vulgare leaves developed at the sun site is reported in Fig. A. The image spatial calibration, performed using a 2 mm micrometer divided into 10 µm units (Reichert-Jung, Cambridge Instruments, Nussloch, Germany), was 0.79 µm per pixel.
The tissue-specific localization of flavonoids and hydroxycinnamates (echinacoside) was estimated by considering that:
(iii) the contribution of flavonoids to the fluorescence signal at 580 nm (F580) differed, for a scale factor α, when cross-sections were excited with UV or blue light. Such changes depended on different intensities of excitation wavelengths as well as on differential absorption coefficients of flavonoids at 365 and 470 nm, respectively:
The distribution of nonflavonoids was then estimated by combining Eqns I–III:
To calculate α we chose a leaf tissue in which flavonoids occurred exclusively. The microspectrofluorometric analysis conclusively allowed calculation of α from the fluorescence signals of the adaxial epidermis of sun leaves (Fig. B). In fact (i) the shape of fluorescence spectra with a negligible contribution of blue-fluorescing compounds under UV excitation, and (ii) the peak of maximum emission at 575 nm under both UV and blue-light excitation, were conclusive for an exclusive accumulation of flavonoids in such a tissue layer (Fig. B). The coefficient α was also estimated by comparing and of a 10−5m ethanol solution of que 3-O-rut with the addition of 0.1% NR. Values of α did not differ appreciably when determined by in vitro or in vivo fluorometric measurements (data not shown).
It has been reported previously that the tissue-specific distribution of blue-green fluorescent dihydroxycinnamates (caffeic and chlorogenic acid, verbascoside and echinacoside; Agati et al., 2002) could conveniently be determined by fluorescence emission at 470 nm of UV-excited cross-sections, as flavonoids did not emit at 470 nm (Schnitzler et al., 1996; Agati et al., 2002). However, we note that (i) wall-bound phenolics, with fluorescence maxima in the 440–480 nm waveband when stained with NR (Strack et al., 1988; Schnitzler et al., 1996), may overlap appreciably with the fluorescence contribution at 470 nm caused by soluble hydroxycinnamates; (ii) blue fluorescence at 460 nm may be partly reabsorbed by chlorophyll in mesophyll cells (Cerovic et al., 1999); and (iii) wall-bound phenolics contribute little to the fluorescence signal at 580 nm (Strack et al., 1988; Agati et al., 2002). As a consequence, the tissue-specific localization of soluble hydroxycinnamates was determined by recording fluorescence signatures in the yellow-red waveband at 580 nm. In particular, profiles of both nonflavonoids and from adaxial epidermal through abaxial epidermal tissues were obtained by plotting the mean fluorescence intensity of each longitudinal row of pixels vs leaf depth (Fig. 6d–f).