Epidermal UV screening and UV absorbance by phenolics
Dissimilar relationships between adaxial and abaxial UV-screening were observed in grape and barley leaves using a modulated UV-A chlorophyll fluorometer: in grape leaves, different radiation treatments produced large variations in epidermal screening but, generally, adaxial UV-A absorbance was twice the corresponding abaxial value (see slope of regression line in Fig. 1a). In contrast, variable UV-A absorbance in barley was similar on both sides of the leaf (Fig. 1b).
Different concentrations of epidermal phenolics are probably involved in the variations in UV screening reported in Fig. 1. To test this, we measured adaxial UV screening in leaf discs using a modulated chlorophyll fluorometer (Xe-PAM) and, subsequently, examined phenolics extracted from these discs by HPLC. Elution profiles were established routinely by absorbance at 314 nm which is the wavelength of the transmission maximum of the UV-B excitation window of our Xe-PAM fluorometer. The substances giving rise to major peaks in chromatograms of grapevine phenolics (Fig. 2a) have been identified earlier (Kolb et al. 2001, 2003). Compounds eluting at retention times below 12 min are hydroxycinnamic acid derivatives with the predominant compound being trans-caffeoyl-tartaric acid, whereas those eluting above 24 min, such as peaks Q1 and Q2, correspond to derivatives of quercetin, a B-ring ortho-di-hydroxylated flavonol, whereas K1 and K2 represent derivatives of kaempferol, a B-ring mono-hydroxylated flavonol.
Phenolics from barley leaves (Fig. 2b) were classified by their retention behaviour and UV absorbance characteristics (Fig. 2d). A distinct group of compounds eluting with short retention times (less than 14 min) possessed spectral properties of hydroxycinnamic acid derivatives from barley leaves (Reuber, Bornman & Weissenböck 1996; Reuber et al. 1997). Further, the conspicuous elution peaks at 7 and 10 min exhibited the absorbance spectrum of pure ferulic acid (see spectrum ‘hca’ in Fig. 2d). Obviously, the group of early eluting compounds are hydroxycinnamic acid derivatives.
In contrast to these hydroxycinnamic acid derivatives, all major compounds eluting beyond 16 min (Fig. 2b) showed an absorbance maximum at around 270 nm and another in the UV-A spectral region (Fig. 2d). This is consistent with the spectral characteristics of flavones which represent the predominant class of flavonoids in barley (Reuber et al. 1997). Barley flavones can be most simply grouped into the B-ring mono-hydroxylated apigenin derivatives, and the B-ring ortho-dihydroxylated luteolin derivatives (Fröst, Harborne & King 1977; Nørbæk, Brandt & Kondo 2000).
We observed two principal types of absorbance spectra of flavones (see Fig. 2d): one category showed a single peak at 270 nm and a long-wavelength maximum at 340 nm, whereas the other exhibited peaks at 255, 270 and 350 nm. It is known, that B-ring-3′,4′ dioxygenated flavones usually exhibit two short-wavelength absorbance maxima but not the mono-oxygenated ones, and, further, that increasing the oxidation of the B-ring in flavones results in a bathochromic shift of the UV-A band (Mabry, Markham & Thomas 1970). Hence, based on absorbance spectra, most of our flavones (see Fig. 2b) could be unequivocally classified as apigenin derivatives (peaks A1 to A4) and luteolin derivatives (L1 to L3). In fact, spectra of L1 to L3 were almost identical to the isoorientin spectrum (luteolin 6-C-glucoside) whereas absorbance spectra and retention behaviour of A1 was identical to saponarin (apigenin 6-C-glucosyl-7-O-glucoside). Compounds L4 and L5 were only tentatively assigned as luteolin derivatives because their UV-A spectra did not match the criteria of B-ring-3′,4′ dioxygenated flavones.
We investigated to what degree variations in UV-shielding of adaxial epidermis can be explained by variations in leaf phenolics by plotting fluorometrically detected UV-B and UV-A absorbance against absorbance of phenolics at 314 and 360 nm, respectively (Fig. 3). Association between data was estimated using linear regression analysis. In grape and in barley leaves, concentrations of the different B-ring ortho-dihydroxylated, and also of the different mono-hydroxylated flavonoids, varied roughly in parallel (data not shown). Therefore, these two groups of compounds were considered separately (Fig. 3e–h).
Figure 3. The four left-handed panels show data from grapevine, other panels represent results from barley. Fluorometrically detected UV-B (AUV-B) and UV-A absorbance (AUV-A) of adaxial leaf sides is plotted against absorbance at 314 nm (A314; panels a, c, e and g) and 360 nm (A360; panels b, d, f and h), respectively, of extracted hydroxycinnamic acids (‘HCA’; panels a to d) and flavonoids (panels e to h). The A314 and A360 are derived from chromatography (see Fig. 2) and are normalized to represent the absorbance that phenolics in the entire leaf blade would have if they were homogeneously dissolved. In panels e and f, the sum of absorbances of the two main kaempferol peaks (K1 and K2 denoted ‘K1,2’) and the two main quercetin peaks (Q1 and Q2 denoted ‘Q1,2’) are considered separately. Likewise, panels g and h treat the sum of absorbances of apigenin (A1 to A4 denoted ‘A1-4’) and luteolin derivatives independently (L1 to L5 denoted ‘L1-5’). Linear regression analyses were applied to all 12 data sets but regression lines are only shown for R2 > 0.1.
Download figure to PowerPoint
Kaempferol and quercetin derivatives in grape leaves but only luteolin derivatives (and not apigenin derivatives) in barley leaves seem to play a role in variable UV-B screening (Fig. 3e–h). In both species, increased epidermal UV-screening was most effectively induced by UV exposure (data not shown). Therefore, the specific increase in luteolin derivatives observed in efficiently screening barley leaves (Fig. 3g & h) corresponds to increased ratios of ortho-dihydroxylated to mono-hydroxylated flavonoids under UV radiation in other species (Markham et al. 1998a, b; Ryan et al. 1998). This supports the notion that UV-induced flavonoid accumulation also increases the leaf's antioxidative capacity because the dihydroxlated flavonoids exhibit higher radical scavenging activity than the mono-hydroxylated compounds (Torel, Cillard & Cillard 1986; Husain, Cillard & Cillard 1987; De Beer et al. 2002; Yamasaki, Sakihama & Ikehara 2003). It remains unexplained why a build-up of high UV screening occurred without a shift to flavonoids with increased B-ring oxidation in grape leaves (Fig. 3e & f) and also in grape berries (Kolb et al. 2003).
The UV absorbance of hydroxycinnamic acids in grape leaves and of apigenin derivatives in barley leaves was always clearly above zero under all experimental conditions (Fig. 3) in agreement with the recent proposal that UV-absorbing compounds in leaves always remain above a certain threshold even under low levels of UV radiation (Liakoura et al. 2003). The present work additionally suggests that, dependent on species, hydroxycinnamic acids or flavonoids can be engaged in maintaining this threshold concentration of UV-B absorbing compounds. The nature of this species-dependency awaits further clarification.
Relationship between epidermal UV absorbance and UV absorbance by phenolics
The previous section demonstrated that different classes of phenolics are involved in variable UV absorbance of the adaxial epidermis in grape and barley leaves. This prompted us to search for species-dependent differences in the relationships between total UV absorbance of leaf phenolics and UV screening of adaxial epidermis. For grape and barley leaves, total phenolic absorbance was derived from the sum of absorbance of all peaks labelled in Fig. 2a and b, respectively: in fact, these peaks represented at least 92% but generally more that 95% of chromatographically determined UV absorbance. This holds for both 314 and 360 nm as the HPLC detection wavelengths; the latter wavelength coincides with the maximum of the excitation window of our Xe-PAM fluorometer used to determine epidermal UV-A screening.
For grape and barley leaves, UV-A and UV-B absorbance of the adaxial epidermis as a function of phenolic absorbance at 360 and 314 nm, respectively, is shown in Fig. 4. In all four plots, epidermal absorbance was considerably smaller than the corresponding phenolic absorbance, except for the smallest values, and curvilinear relationships were observed. Curvilinear behaviour was already apparent in Fig. 3 although these data were analysed using linear regression; for example, outermost data points of absorbance by quercetin derivatives versus epidermal UV-A absorbance in grape leaves fell below the regression line but the greater proportion of central points resided above the line (Fig. 3f). Curvilinear relations differed markedly between species: initial slopes determined by linear regression for A314 or A360 < 2 were about 0.45 for grape leaves (Fig. 4a & c) but 0.25 for barley leaves (Fig. 4b & d).
Figure 4. Fluorometrically determined UV absorbance of adaxial sides of grape (a and c) and barley leaves (b and d) as a function of total UV absorbance of all major extractable phenolics. In (a) and (b), epidermal UV-A absorbance is plotted versus phenolic absorbance at 360 nm; epidermal UV-B screening versus phenolic absorbance at 314 nm is depicted in (c) and (d). Curved lines correspond to the best fit of our model function (Eqn 4) to the data. Parameters of these curves are summarized in Fig. 5.
Download figure to PowerPoint
As stated in Material and Methods, stray UV radiation from our leaf samples could have influenced calculations of epidermal transmittance particularly when true epidermal UV screening results in low intensities of FUV-B or FUV-A. Hence, curved relationships in Fig. 4 could have been the result of measuring artefacts. However, in the present study we corrected for stray radiation by employing a method which obtains realistic estimates for background signals in Xe-PAM measurements (see Material and Methods). Further, curvilinear relationships obtained with grape leaves were still apparent when the background signals from an aluminum foil, which certainly are higher than those from leaf samples, were subtracted from FUV-B and FUV-A measurements (not shown). Therefore, it seems unlikely that the curvilinear behaviour of data in Fig. 4 originates from instrumental artifacts, and we looked for a model function which explains these relationships by taking into account differences in the optical properties of phenolics in the leaf and in solution.
The lowest absorbance of the adaxial epidermis is observed for lowest absorbance of extracted phenolics (Fig. 4). In this case, our conception requires that TUVv equals zero and minimum concentrations of UV-absorbing phenolics are present in vacuoles. We assume for lowest epidermal absorbance that transmittance of the epidermal area including vacuoules (aVc) does not deviate markedly from transmittance of the remaining epidermal area (1 −aVc), and consider the epidermis as a homogeneous optical layer characterized by basal transmittance TUVbasal. Considering variable UV screening, the product of TUVv and TUVbasal determines epidermal transmittance of aVc; but, TUVbasal alone describes transmittance of (1 − aVc). Transmittance of the entire epidermis, TUV, corresponds to the sum of (aVc·TUVv·TUVbasal) plus [(1 − aVc)·TUVbasal]
Replacing transmittance by absorbance terms yields
We further define the absorbance of extracted phenolics, AEX.UV, which includes the absorbance of phenolics that are present when epidermal screening is minimal, AEX.UVbasal, plus variable phenolic absorbance, AUVv. Note that AEX.UVbasal solely results from extracted phenolics but AUVbasal is composed of soluble and insoluble phenolics as well as other epidermal constituents absorbing in the UV. We assume that during our relatively short exposure intervals, changes in AUVbasal are negligible in comparison with changes in variable absorbance of the adaxial epidermal vacuole (AUVv). As explained above, AUVv results from variable phenolic absorbance located in these vacuoles. Extraction distributes the vacuole-confined absorbance homogeneously and, thus, dilutes AUVv by the factor aVc. Figure 1 indicates that phenolic absorbance in abaxial sides change in parallel to that of the adaxial side. Hence, variable phenolics from adaxial side (aVc·AUVv) is multiplied by (1 + f) to obtain the entire variable absorbance in the extract. More generally, the f-value is defined as the phenolic absorbance not involved in adaxial UV screening but varying parallel to phenolic absorbance active in adaxial UV screening, expressed relative to phenolic absorbance active in adaxial UV screening. One can formulate
Combining Eqns 2 and 3 results in a model function that describes the four data sets in Fig. 4, where the AUV corresponds to ordinates and the AEX.UV to the abscissas
For each species and UV range, the unknown parameters AUVbasal and AEX.UVbasal were assessed from means of the five lowest values of epidermal absorbance and absorbance of extracted phenolics, respectively. In all cases, these minimum values originated from the group of plants grown under shaded and practically UV-free growth conditions. Using the mean values for AEX.UVbasal and AUVbasal, the free parameters, aVc and f, were estimated by fitting our model function to the four data sets in Fig. 4. Solid lines in Fig. 4 represent best fits of theory to experiment: corresponding parameters are depicted in Fig. 5.
Figure 5. Parameters of the four model functions (solid lines in Fig. 4). ‘UV-A’ and ‘UV-B’ refers to curves in the UV-A (Fig. 4a & b) and the UV-B spectral range (Fig. 4c & d), respectively. Data for grape leaves are shown as open and those for barley leaves as closed bars. (a) and (b) depict basal absorbance of extracted phenolics and basal absorbance of adaxial epidermis, respectively, obtained with leaves grown in a shaded greenhouse and in the virtual absence of UV radiation (error bars: standard error, n = 5). Other panels depict parameter estimates obtained by fitting our model function to experimental data; error bars represent standard errors calculated by the curve fitting program. (c) Relative area of adaxial epidermis enclosing vacuoles. (d) UV absorbance of extracted phenolics which are not involved in adaxial UV screening but vary parallel to UV absorbance of phenolics active in adaxial UV screening, expressed relative to UV absorbance of the adaxially screening phenolics.
Download figure to PowerPoint
Basal epidermal UV absorbance was always higher than 0.25 (Fig. 5b). In comparison, we calculated from the transmittance spectrum of a methanol-extracted epidermis from Vicia faba (Markstädter et al. 2001) values of 0.07 and 0.01 at 314 and 360 nm, respectively, using the absorbance at 490 nm as the baseline (the 490 nm represents the wavelength position of maximum blue-green excitation of our Xe-PAM fluorometer). The much higher UV screening of the intact epidermis is consistent with presence of some UV-absorbing phenolics in the epidermis even in the absence of UV stress (Liakoura et al. 2003).
Direct comparison of basal absorbance of the epidermis (Fig. 5b) with basal absorbance of extracted phenolics (Fig. 5a) is difficult because the former also includes contributions by insoluble phenolics and non-phenolic factors. However, in grapevine, basal epidermal screening in the UV-B was higher than in the UV-A, and this pattern was paralleled by basal absorbance of extracted phenolics. The higher UV–B absorbance in these extracts is consistent with considerable amounts of the preferably UV-B-absorbing (Fig. 2c) hydroxycinnamic acids in grape leaves exhibiting basal epidermal UV screening (Fig. 3a). That basal epidermal screening in the UV-B was higher than in the UV-A is therefore well explained by localization of a major part of hydroxycinnamic acids in the adaxial epidermis. Similar basal epidermal screening for both UV spectral ranges in barley leaves (Fig. 5b) agrees with the presence of apigenin derivatives in the epidermis (Fig. 3), which absorb in the UV-A as well as in the UV-B spectral region (Fig. 2d).
When variable UV screening is considered, it is important to realize that our model function (Eqn 4) describes a straight line for aVc = 1; that is when spaces between strongly UV-absorbing vacuoles are negligibly narrow. Fitting theory to experiment, however, always yielded curvilinear relations with estimates for aVc of around 0.9 (Fig. 5c). These data agree with Day et al. (1993) who have demonstrated that the epidermis of outdoor-grown leaves behave as an inhomogeneous UV filter exhibiting an ‘optical sieve effect’ due to UV-transparent gaps between highly UV-absorbing vacuoles.
Unlike aVc, the f-value differs considerably between species (Fig. 5d): In grape leaves, the f-value was 0.75 and 0.5 for the UV-A and UV-B range, respectively; but the comparable values were 2.3 and 1.7 in barley leaves. In grape leaves, adaxial UV-A absorbance was about twice as high as the abaxial one (Fig. 1a) which corresponds to an f-value of 0.5. In comparison, similar epidermal UV-A absorbance of both barley leaf sides results in an f-value of approximately 1 (Fig. 1b). Therefore, in grape leaves, the theoretical f-values were 150 and 100% in the UV-A and UV-B range, respectively, of the f-value derived from Fig. 1, but the equivalent percentages were 230 and 170% in barley. An obvious explanation for the markedly higher deviations of theory from experiment in barley is that a considerable fraction of variable phenolic absorbance is located in the barley leaf mesophyll but not in the grapevine leaf mesophyll. This agrees well with Liu et al. (1995) who deduced from their data that epidermal and mesophyll flavonoids in barley increased in parallel in response to artificial UV-B radiation. Evidently, different allocation patterns of phenolics gave rise to the much different initial slopes of the relationships between adaxial UV absorbance and UV absorbance of leaf phenolics observed for grape and barley leaves (Fig. 4).