Metabolomic and physiological responses reveal multi-phasic acclimation of Arabidopsis thaliana to chronic UV radiation


J. A. Lake. Fax: +44 114 222 0002; e-mail:


Biochemical changes in vivo and pathway interactions were investigated using integrated physiological and metabolic responses of Arabidopsis thaliana L. to ultraviolet (UV) radiation (280–400 nm) at 9.96 kJ m−2 d−1 over the entire life cycle from seed to seed (8 weeks). Columbia-0 (Col-0) and a UV-B sensitive accession (fah-1) showed significant (P < 0.001) reductions in leaf growth after 6 weeks. Col-0 recovered growth after 8 weeks, with recovery corresponding to a switch from production of phenylpropanoids to flavonoids. fah-1 failed to recover, indicating that sinapate production is an essential component of recovery. Epidermal features show that UV radiation caused significant (P < 0.001) increases in trichome density, which may act as a structural defence response. Stomatal indices showed a significant (P < 0.0001) reduction in Col-0 and a significant (P < 0.001) increase in fah-1. Epidermal cell density was significantly increased under UV radiation on the abaxial leaf surface, suggesting that that a fully functioning phenylpropanoid pathway is a requirement for cell expansion and leaf development. Despite wild-type acclimation, the costs of adaptation lead to reduced plant fitness by decreasing flower numbers and total seed biomass. A multi-phasic acclimation to UV radiation and the induction of specific metabolites link stress-induced biochemical responses to enhanced acclimation.


Solar radiation provides the energy driving photosynthetic carbon fixation in plants. Ultraviolet (UV)-A radiation (λ = 315–400 nm) represents 6.3% and UV-B (λ = 280–315 nm) represents 1.5% of incoming solar radiation (Hollosy 2002). Stratospheric ozone loss may cause relatively large increases in the amount of UV-B radiation reaching the earth's surface but with virtually no effect on levels of UV-A (Hollosy 2002). Although both UV-A and UV-B are biologically harmful to terrestrial plants, the excess UV-B component penetrating the stratospheric ozone layer can induce DNA damage by dimerization of thymidine residues and membrane damage by lipid and protein oxidation (Rozema et al. 1997; Shanker 2006). Consequently, plants face a trade-off between the requirement for irradiance and managing the detrimental effects of UV radiation. This dilemma may have been particularly acute in the geological past when atmospheric chemistry model calculations indicate that the stability of the stratospheric ozone layer shielding Earth's surface biota from excess UV-B radiation can be disrupted by geological phenomena (Lamarque et al. 2006; Beerling et al. 2007; Lamarque, Kiehl & Orlando 2007).

Stratospheric ozone depletion due to anthropogenic chlorofluorocarbon (CFC) emissions from the early 1970s (Farman, Gardiner & Shanklin 1985) has also increased UV-B radiation exposure of high southern (Newsham et al. 2005) and northern latitude (Weatherhead et al. 2005) ecosystems, with negative impacts on photosynthetic primary production (Searles, Flint & Caldwell 2001). Despite reductions in CFC emissions due to the implementation of the Montreal Protocol (WMO 2003), models project continued severe springtime enhancement of surface UV-B by 2020, relative to 1979–1992 conditions due to the long lifetime of CFCs (several hundred years), the continued release of halogenated compounds and changes in stability of the Arctic and Antarctic polar vortices due to climate change (Taalas et al. 2000). These processes will increase the vulnerability of the stratospheric ozone column (Taalas et al. 2000; Butchart & Scalfe 2001), although by 2050, increases in the rate of transportation of air from the troposphere to the stratosphere may result in enhanced removal of CFCs (Butchart & Scalfe 2001).

Plants respond to UV exposure in two fundamental ways: through the production of UV-absorbing compounds and by DNA repair (Jansen, Gaba & Greenberg 1998; Tuteja et al. 2001). These primary, fundamental responses ultimately impact on tertiary responses such as growth, as a consequence of the costs of protection measures. Enhancements in surface UV-B radiation (∼12% –McKenzie, Connor & Bodeker 1999), for example, stimulate the synthesis of UV-B absorbing phenolic compounds and their deposition around photosynthetic tissues, as well as in the walls of pollen spores (Rozema et al. 2001, 2002; Lomax et al. 2008). Nevertheless, short-term UV-B exposure alters the morphology and physiology of plants by reducing leaf area (Ren et al. 2006), root and hypocotyl length, epidermal cell expansion (Kim, Tennessen & Last 1998) and increasing trichome density (TD) (Tattini et al. 2000; Manetas 2003). It can also reduce (Booij-James et al. 2000) or increase (Yang & Yao 2008) photosynthetic rate (A), exert complex changes in stomatal behaviour (gs) (Jansen & van den Noort 2000) and reduce plant fitness (Conner & Neumeier 2002; Feng et al. 2007). Studies of UV-A radiation rely on ambient UV radiation, exclusion of UV-B and total UV radiation to infer ambient UV-A specific responses (Kotilainen et al. 2008; Yang & Yao 2008). Many responses are similar to UV-B, but are less intense.

The majority of UV-related studies have investigated the responses of plants to UV-B doses commensurate with anthropogenic ozone loss, for example, maximum doses of 5.8 kJ m−2 d−1 biological equivalent (BE) to simulate a 15% reduction in stratospheric ozone under clear sky conditions at Abisko, Sweden (Phoenix et al. 2001). However, the physiological and morphological responses of plants to prolonged doses of UV-B (>20 kJ m−2 d−1 BE) (Ries et al. 2000) exceeding those caused by anthropogenic ozone depletion may also be important on evolutionary timescales (Visscher et al. 2004; Lamarque et al. 2006, 2007; Beerling et al. 2007).

Here, we report the findings of a study designed to investigate the effects of prolonged (chronic) elevated UV radiation dosage of 9.96 kJ m−2 d−1 applied over the entire life cycle (8 weeks) of two accessions of Arabidopsis thaliana L., Columbia-0 (Col-0) and the UV-B sensitive mutant (fah-1) in the Col-0 background (Chapple et al. 1992). These accessions were selected to elucidate the biochemical changes involved in acclimation processes in vivo and test the hypotheses set out below. The fah-1 accession is UV-B sensitive because of a mutation in the phenylpropanoid pathway resulting in an inability to synthesize the enzyme ferulate-5-hydroxylase responsible for converting ferulic acid to sinapic acid. We aimed to characterize the morphological, physiological and growth responses of the model plant A. thaliana to chronic high UV exposure. Determination of the metabolic costs of UV protection on plant fitness, defined in terms of flower, seed production and seed viability, allows assessment of the extent to which UV radiation is an important selective pressure on evolutionary timescales.

Our experiments were designed to test the following hypotheses:

  • 1Chronic exposure to UV radiation results in altered epidermal morphology consistent with the effect of short-term acute dose studies.
  • 2The lack of a fully functioning phenylpropanoid pathway has severe effects on leaf development and growth under exposure to UV radiation incurring high metabolic costs.
  • 3Such costs of UV protection mechanisms will significantly impact on plant fitness.


Plant material

Arabidopsis thaliana Col-0 and the UV-B sensitive mutant (fah-1 in the Col-0 background) (Nottingham Arabidopsis Stock Centre, Nottingham, UK) were sown onto multi-purpose compost (Arthur Bowyers, William Sinclair Horticulture, Lincoln, UK) covered with plastic film and stratified for 3 d at 4 °C. They were then transferred into controlled environment cabinets (Sanyo-Gallenkamp Model: P61700H/FM/RO-HQI, Sanyo-Gallenkamp, Loughborough, UK) and grown under a day/night regime of 8/16 h at 21/15 °C, 60% ± 5 relative humidity (RH). Photosynthetically active radiation (PAR = λ 400–700 nm) at 250 µmol m−2 s−1 at plant rosette height. UV light was supplied by a bank of nine Sanolux UV high pressure lamps (323 1219 – Phillips, Guildford, UK, in an array recommended by the manufacturer to produce optimal spacing) increasing UV-B levels to 9.96 ± 0.01 kJ m−2 d−1 at plant rosette height over the entire day length and excluded in the control cabinet using specific UV cut-off wavelength filters (226 Lee U.V., Lee Filters, Andover, UK – Supporting Information Fig. S1). UV-B levels were measured using a Sola-Check System (Solatell, Croyden, UK) cross-calibrated with a USB2000+ spectroradiometer (Ocean Optics via Photonic Solutions PLC, Edinburgh, UK). Analyses were carried out at both the 6 and 8 week stage, unless otherwise stated. For experimental conditions, see Supporting Information.

Leaf area and morphology

At 6 and 8 weeks, mature rosette leaves complete with their petioles were collected from different individuals and measured using a leaf area meter (DeltaT Area Meter, Delta T Devices, Cambridge, UK). Three fully mature leaves from nine different plants in each treatment were measured at each time point.

Leaves where first viewed directly under a dissecting microscope (73082, Nikon, Tokyo, Japan) at ×30 and TD; the number of trichomes per mm2 for the adaxial (upper) leaf surface was determined. From this and epidermal cell density (ED) data, the trichome index (TI) was calculated as the ratio of trichomes to epidermal cells (as determined below). This measure is independent of leaf area and analogous to stomatal index (SI). The same leaves that were used for trichome analysis were also used for stomatal analysis. Epidermal features were determined from surface-activated dental putty (Coltene Whaledent, Altstatten, Switzerland) impressions of both leaf surfaces. Cellulose varnish was then applied to the dental impression, mounted on slides and viewed through a light microscope (CX40RF200, Olympus, Watford, UK) at ×400. Ten fields of view of 3 leaves per plant from 5 individual plants per treatment were analysed for stomatal density (SD), SI and ED (Salisbury 1928). Guard cell length (GCL) was measured using image analysis software (Quantimet 500, Leitz, UK) coupled to a light microscope (Leitz Laborlux S, Leitz). Ten randomly sampled stomata on 3 leaves per plant from each treatment were measured on both leaf surfaces. Statistical analyses utilized a two-way repeated measures analysis of variance (Minitab v12, Coventry, UK) (Table 1).

Table 1.  Results of two-way repeated-measures analysis of variance for leaf morphological traits and carbon isotope analysis
Treatment, accession and time interactions (Col-0, fah-1)
 UVspeciesUV × speciesTimeUV × timeSpecies × time
Leaf area66.1<0.00125.2<0.00126.5<0.00112.7<0.0010.50.462.60.105
Treatment and accession interactions (Col-0, fah-1)Treatment and time interactions (Col-0)
  1. The two-way component tested effects of UV-B treatment and accession (Col-0, fah-1), whereas the repeated measures component examined the effect of time (measured at 6 or 8 weeks growth) for Col-0.

  2. Col-0, Columbia-0; UV, ultraviolet; SD, stomatal density; SI, stomatal index; ED, epidermal cell density; TD, trichome density; TI, trichome index.

 ED 33.8<0.001269<0.00199.3<0.00133.1<0.00140.1<0.0015.8<0.05

Leaf gas exchange and carbon isotope analysis

Leaf gas exchange of Col-0 plants at the 6 week stage were measured using a CIRAS-1 infrared gas analyser (PP Systems, Amesbury, MA, USA) using a broadleaf plant leaf chamber (PLC). Leaf area was determined by placing a transparent sheet over the PLC and tracing the leaf contour visible through the PLC window after steady state measurements were recorded. Mature rosette leaves were placed in the portable leaf cuvette directly outside the growth cabinet (high UV levels in the treatment cabinet negated in situ measurement) until steady state values were obtained. Gas exchange analysis was not possible on the fah-1 mutants, as the plants were too small. Temperature and RH in the cuvette were matched to daytime growth conditions.

At the 6 week stage, 0.1 mg of leaf material per plant was collected and dried for 1 week at 70 °C for carbon isotope analysis. Five plants of each ecotype from each treatment were analysed. Samples were ground in a pestle and mortar. Measurements were made using an ANCA GSL preparation module, coupled to a 20–20 stable isotope analyser (PDZ Europa, Cheshire, UK) Air samples from growth cabinets were pumped into 10 mL evacuated gas tight vials (Labco Exetainer Vials, Labco Ltd, High Wycombe, UK) and analysed on the same analyser. Water use efficiency (WUE-Δ) and transpiration (A/E) are theoretically linked by the equations:


where a = diffusional fractionation in air = −4.4‰, b = discrimination by ribulose 1·5-bisphosphate carboxylase/oxygenase = −27 to −30‰, Ci and Ca = intercellular and ambient pCO2, respectively, (Farquhar, Ehleringer & Hubick 1989). Discrimination (taken to represent a measure of WUE) is calculated as:


where δ13Ca is the isotopic composition of the air inside growth cabinet, δ13Cp is the isotopic composition of the leaf material relative to the PDB (Pee Dee Belemnite) standard (Farquhar et al. 1989). Statistical analyses utilized a Student's t-test, two-sample means (Minitab v12).

Floral and seed production measurements

Following onset of flowering in Col-0, number of flowers, number of open flowers, number of pollinated flowers and floral stem height were obtained for each of 20 plants on a weekly basis for 3 weeks. Pollinated flowers were those exhibiting a long stamen exiting the flower and siliques were included in this count. Siliques were harvested and dried at 60 °C for 72 h, the seeds removed and weighed. Twenty-five seeds from each plant were placed onto damp filter paper in a Petri dish and stratified for 2 d at 4 °C. Petri dishes were placed into a controlled environment growth cabinet (Sanyo-Gallenkamp Model: P61700H/FM/RO-HQI, UK) and the number of germinated seeds per plant were recorded over the next 10 d. UV-B levels at the top of flowering stems were higher than rosette levels as they were 20 cm closer to light source. This is calculated as 1.72 times the level at rosette height (17.1 kJ m−2 d−1) using the inverse square law. Statistical analyses utilized Student's t-test, two-sample means (Minitab v12).

Metabolomic analysis

Metabolomic extracts were obtained from leaves of both accessions collected at 6 weeks and for Col-0 at 8 weeks (insufficient material for analysis of fah-1 at 8 weeks). Leaves were immediately immersed into liquid nitrogen and stored at −80 °C. Extractions were taken using the cold methanol/chloroform/water method (Overy et al. 2005; Davey et al. 2008). In addition, 250 µL sample was diluted in 500 µL 1:1 methanol : water and directly injected into an electrospray-ionization time-of-flight mass spectrometer (ESI-ToF-MS; LCT. Micromass, Manchester, UK) operating in positive and negative ionization mode as detailed in Overy et al. (2005). Spectra were obtained from 50 to 1000 m/z. Data were processed into 1 Dalton (Da) bins using the method of Overy et al. (2005). Binned m/z and total ion count (TIO) values were explored by principal component analysis (PCA) using Simca-P V.11.0 (Umetrics, Umea, Sweden) using 1 Da binned masses (m/z) as the primary variable and populations as the observation variable. R2 (correlation coefficient) values from the analysis were obtained for each m/z to explain how each primary variable (m/z value) contributed to the discrimination of observation variables in the PCA. m/z values with a R2 close to 1.0 were investigated further using the online databases KEGG, Aracyt and the Dictionary of Natural Products for putative compound identification.

Flavonoid and phenylpropanoid identification and quantification

Leaf metabolite extractions were analysed using high-performance liquid chromatography (HPLC – Hewlett Packard 1090 – series 2). Samples were separated on a Phenomenex Gemini column (52 × 2.0 mm) using acetonitrile (solvent B) and 0.1% formic acid (solvent A), applying a 43 min gradient of B increasing so that initial A : B (80:20 v/v); 2 min (80:20); 30 min (60:40) at a flow rate of 0.1 mL min−1, then 35 min (0:100); 38 min (80:20), 43 min (80:20) at a flow rate of 0.25 mL min−1 as in Davey et al. (2004). Compounds eluting from the column were passed through a photodiode array detector (PDAD) and monitored for UV absorbance between 200 nm and 500 nm. Traces were visualized at 287 nm and quantification was achieved using sinapic and ferulic acid standard curves (0–20 µg mL−1). The traces showed absorbances in the esterified form so quantification was based on free sinapic acid or ferulic acid equivalents. Quercetin and kaempferol were quantified using the sinapic acid standard curve; therefore, these compounds are expressed as a sinapic acid equivalence basis. Elutions containing compounds of interest were collected in 1.5 mL eppendorf tubes and stored at −80 °C until mass identification and fragmentation by mass spectrometry.

Flavonoid glycosides and phenylpropanoid esters were identified using an atmospheric–pressure–ionization (API) triple quadruple mass spectrometer (PE SCIEX API 3, Applied Biosystems, Westborough, MA, USA) in the positive and negative ionization mode (ionization voltage: −3.8 kV; interface plate voltage: −400; orifice plate voltage: −55). Data were acquired over 50–350 m/z or 100–600 m/z. Crude samples and HPLC eluants that contained flavonoids and phenylpropanoids collected from the HPLC were injected at 10 µL min−1 and analysed. Parent ions of compounds of interest were detected and fragmented by collisionally activated dissociation (CAD) using nitrogen gas to identify daughter ion fragments of the selected parent compound. Identification of phenolic glycoside conjugates is in agreement with Stobiecki et al. (2006). Statistical analysis of compound concentration utilized Student's t-test, two-sample means (Minitab v12).


Leaf area and morphology

The individual leaf area of mature rosette leaves grown under elevated UV was less than that grown in the absence of UV in both Col-0 and fah-1 after 6 weeks growth, with rosette leaf area of the mutant being significantly less than the wild type following UV treatment (Fig. 1a). However, after 8 weeks (Fig. 1d), the wild-type leaf area was larger than that of the control plants (Fig. 1c), whereas the mutant failed to recover (Fig. 1f). Leaf physiognomy was also affected by UV radiation with exposed plants, particularly the mutant, having a reduced petiole, a more compact rosette and consequently a higher degree of self-shading (Fig. 1b–f).

Figure 1.

Changes in Arabidopsis thaliana[Columbia-0 (Col-0) and fah-1] leaf morphology and physiognomy on exposure to +ultraviolet (UV) and −UV radiation. Mean leaf area at (a) 6 weeks and (d) 8 weeks. Photographed at 6 weeks [(b) Col-0, (e) fah-1] and at 8 weeks [(c) Col-0, (f) fah-1]. Scale bar is 1 cm (Different letters denote statistical significance where ≤ 0.05 Tukey pairwise comparisons of accession and treatment. n = 9, bar = SE mean).

After 6 weeks, TD was significantly higher for both accessions grown under high UV (Fig. 2a) with fah-1 showing significantly greater numbers of trichomes per unit area than the wild type. TI, a leaf area independent measure, showed that Col-0 maintained a slightly higher number of trichomes per epidermal cell following treatment, compared with the mutant (Fig. 2e).

Figure 2.

Epidermal morphology of Arabidopsis thaliana[Columbia-0 (Col-0) and fah-1] exposed to +ultraviolet (UV) and −UV radiation. Trichome density (a) and trichome index (e) of the adaxial surface, stomatal density (b) and (f), stomatal indices (c) and (g), epidermal cell density (d) and (h) of the abaxial and adaxial leaf surfaces, respectively (Different letters denote statistical significance where ≤ 0.05 Tukey pairwise comparisons of accession and treatment. n = 27, bar = SE mean). Measured at 6 weeks.

Col-0 SD on both the abaxial (Fig. 2b) and adaxial (Fig. 2f) leaf surface decreased dramatically under high UV, 30 and 40%, respectively, compared to controls. SI (Fig. 2c,g) also decreased by 14 and 9% for the respective surfaces. ED decreased by 10% on the abaxial (Fig. 2d) but without significant change on the adaxial leaf surface (Fig. 2h).

The mutant exhibited different responses to the wild type on both abaxial and adaxial surfaces for SD, showing a 35 and 16% increase, respectively (Fig. 2b,f). SI increased by 11% on the abaxial surface but was not significantly reduced on the adaxial surface (Fig. 2c,g). Effects on ED show a larger response than the wild type, increasing by ∼30% on both surfaces (Fig. 2d,h). GCL did not differ for Col-0 or the mutant on the abaxial surface but did significantly increase on the adaxial surface for both Col-0 (mean ± SE mean of 21.6 ± 0.4, to 23.1 ± 0.5) and fah-1 (18.6 ± 0.5, to 21.1 ± 0.6), with the wild type having significantly larger guard cells than the mutant when grown under +UV.

Leaf gas exchange and C-isotope responses

Photosynthetic rate (A) was unaltered in Col-0 after 6 weeks growth under +UV relative to control plants. However, transpiration rate (E) and stomatal conductance (gs) increased significantly under +UV conditions (Fig. 3a). Carbon isotope discrimination (Δ13C) also increased significantly for both Col-0 and fah-1 with UV treatment which corresponds with a rise in Ci/Ca, and is consistent with high UV plants maintaining higher operational gs values (Fig. 3b).

Figure 3.

Physiological measurements of Columbia-0 (Col-0) exposed to +ultraviolet (UV) and −UV for 6 weeks. (a) Photosynthetic rate, A (µmol m−2 s−1), evapotranspiration, E (mol m−2 s−1), stomatal conductance, gs (mmol m−2 s−1). (b) Carbon isotope discrimination (Δ13C) for both Col-0 and fah-1. [(a) Student's t-test *** = ≤ 0.0005, * = ≤ 0.05, ns, not significant. n = 5, bar = SE mean. (b) Different letters denote statistical significance where ≤ 0.05 Tukey pairwise comparisons of accession and treatment. n = 5, bar = SE mean].

As GCL on the adaxial leaf surface is significantly larger in both Col-0 and fah-1 in enhanced UV conditions, and despite a reduction in stomatal numbers, this may allow both transpiration rate (E) and stomatal conductance (gs) to remain significantly higher under UV (Fig. 3a).


PCA (Fig. 4) illustrates differences in metabolomic fingerprints of both treatment and accession (Col-0 and fah-1) after 6 weeks growth. Clear separation between Col-0 (Fig. 4a) and fah-1 (Fig. 4b) under elevated UV and non-UV treatments and between accessions (Fig. 4c) reflects changes of peak intensity of metabolites within individual spectra. Significant differences in the percentage TIC (%TIC) of four of the most discriminating m/z− values were identified by PCA analyses (Fig. 5); 223 and 339 in Col-0 (Fig. 5a) and 193 and 309 in fah-1 (Fig. 5c).

Figure 4.

Principal component analysis (PCA) analyses of metabolomic fingerprints of plants grown under +ultraviolet (UV) and −UV treatments for 6 weeks in negative ionization mode. PCA1 versus PCA2 show separation between treatments in Columbia-0 (Col-0) (a) and fah-1 (b). PCA1 versus PCA3 (c) shows separation between the wild type and mutant. Each datum point represents an individual leaf used for analysis (n = 15).

Figure 5.

Relative compound abundance and compound identification for Columbia-0 (Col-0) (a) and fah-1 (c) where %TIC relates to total ion count within sample in negative ionization mode. m/z− 223 is sinapic acid; 339 is sinapoyl malate; 193 is ferulic acid and 309 is feruloyl malate. Typical high-performance liquid chromatography chromatograms of shoot extract from Col-0 (b) and fah-1 (d) exposed to +ultraviolet (UV) or −UV for 6 weeks. Peaks were identified as quercetin glycoside (Q); kaempferol glycoside (K); sinapoyl malate (S) and feruloyl malate (F). (Statistical analyses show difference between treatments Student's t-tests; *** = ≤ 0.0005. n = 5, bar = SE mean).

HPLC analysis identified eight UV-absorbing compounds (Fig. 5b,d). The UV absorbance spectra of four unknown peaks (retention times at 12.7, 15.3, 18.4 and 18.6 min) showed close resemblance to the phenylpropanoid and flavonol standards (summarized in Table 2). Shifts in the band I λ max values, compared to aglyca standards, indicated the presence of glycosidic side groups. The chromatography and spiked samples and standards showed that the peaks were not sinapic and ferulic in the free form but were likely to be conjugated as indicated by a shift in the sample retention time.

Table 2.  Identification of phenolic metabolites of Arabidopsis thaliana compounds by molecular mass ionisation (m/z-), MS/MS and UV-spectral maxima (PDA)
LC Rt (min)Compound identification and accurate massPDA
UV λmax (nm) of band I and band II in the phenolic moiety
Reference PDA UV λmax (nm) of band I and band II in the phenolic moietyObtained parent and fragment molecular mass ions from LCT m/z-Reference mass ion on LCT m/z-Obtained parent and fragment molecular mass ions from Sciex m/z- or m/z+Reference mass and fragment ions on Sciex m/z-
  1. PDA, photodiode array detector; UV, ultraviolet.

12.7Quercetin Diglycoside
Quercetin 3-rhamnoside (Q-3-R) = 448.10
Quercetin = 302.04
350.5, 266.5348.5, 264.5 (Q-3-R)592.41, 446.72, 285.02, 147.19446.84, 300.08, 147.22 (Q-3-R)592.74, 446.85, 285.36+446.88, 301.03, 146.8 (Q-3-R)
15.3Kaempferol 4′,7-dirhamnoside = 578.16
Kaempferol (K) = 286.05
Kaempferol 4′-rhamnoside = 432.11
348.5, 266.5366.5, 266.5 (K)576.59, 430.85, 285.10, 147.22285.06 (K)576.95, 429.0, 286.5+
285.09 (K)
18.4Sinapoyl malate = 340.07
Sinapic acid (S) = 224.07
328.5, 238.5324.5, 238.5 (S)339.02
223.20, 116.30
223.23, 116.30 (S)338.64, 222.63, 116.08222.76 (S)
18.6Feruloyl malate = 310.07
Ferulic acid (F) = 194.06
326.5, 234.5322.5, 238.5 (F)309.10, 193.07193.07 (F)308.92, 192.96, 115.2192.92 (F)

Compound identification by MS–MS of the crude sample and the four UV-absorbing peaks of interest from the HPLC fractions showed that parent and fragment masses matched the putatively identified compounds from the UV traces (Table 2). Fragmentation of the main component of the eluant at 12.7 min (m/z− 593) showed prominent peaks occurring at a m/z− of 447 (corresponding to quercetin-3-rhamnoside m/z− 448), m/z− of 147 (potentially a glycoside) and a m/z− of 301 (quercetin m/z− 301). This fragmentation pattern also suggested the presence of a quercetin di-glycoside within this fraction. MS–MS analysis of the HPLC eluant at 15.3 min identified a prominent peak present at a m/z− of 576 and another m/z− at 430. These mass numbers match those of a kaempferol di-glycoside and kaempferol glycoside with kaempferol at a m/z− of 285. When the parent ion (m/z− 576) was fragmented, spectra showed peaks at a m/z− of 430 and 147. Under further ionization, spectra consistently showed peaks present at a m/z− of 430, 285 and 147. These mass numbers correspond to a kaempferol di-glycoside being broken into a kaempferol glycoside, free kaempferol (m/z− 285) and the free glycoside (m/z− 147). The parent compound at a m/z− of 339 was detected in the 18.4 min fraction and produced fragment ions at a m/z− of 115 and 223, indicating the presence of sinapoyl malate (the m/z− of malate being 115 and that of sinapate being 223). Similarly, the compound with a m/z− at 309, and detected in the 18.6 min fraction, produced fragment ions at a m/z− of 115 and 193, indicating the presence of feruloyl malate (m/z− of 115, again representing malate and the m/z− of 193 being ferulic acid).

Ferulic acid was detected in fah-1 but not in Col-0. Quantification of feruloyl malate by HPLC showed that fah-1 produced significantly increased ferulic acid concentrations when grown under high UV conditions for 6 weeks, compared to an undetectable amount in control plants (Fig. 6a). Metabolomic data for fah-1 were not available for 8 weeks.

Figure 6.

Quantification of the major identified metabolites in fah-1 after 6 weeks (a) and in Columbia-0 (Col-0) after 6 and 8 weeks (b) exposure to either +ultraviolet (UV) or −UV radiation. Sinapoyl malate (S) and feruloyl malate (F) are expressed as sinapic acid and ferulic acid equivalents respectively. Kaempferol (K) and quercetin (Q) glycosides are expressed as sinapic acid equivalents. (Student's t-tests; *** = ≤ 0.0005, ** = ≤ 0.005, * = P ≤ 0.05, ns, not significant. n = 3, bar = SE mean). fwt, fresh weight.

Sinapoyl malate was detected in Col-0 but not in fah-1. Quantification of sinapoyl malate by HPLC showed a significant 2.8-fold increase in Col-0 tissue following exposure to chronic UV for 6 weeks (Fig. 6b). At 8 weeks exposure, the synapoyl malate concentration in UV-treated plants remained high with control plants increasing levels of this compound comparable to those in UV-exposed plants (Fig. 6b).

There were significantly increased levels of quercetin–glycoside and kaempferol–glycoside in both accessions when exposed to UV compared with −UV for 6 weeks (Fig. 6). In Col-0, at 8 weeks exposure, levels of quercetin decreased when compared with those exposed for 6 weeks. However, kaempferol concentrations remained significantly higher than plants grown without UV (Fig. 6b).

Floral and seed production

In Col-0, floral initiation was delayed under elevated UV (Fig. 7a), which resulted in a 54% reduction in the number of flowers produced (Fig. 7b). Mean seed weight was significantly reduced by 46% under UV treatment compared with −UV plants (Fig. 7c) Even though there was a reduction in mean germination rate, this was not significantly different between treatments (Fig. 7d). fah-1 plants failed to flower over the duration of the experiment.

Figure 7.

Floral characteristics of Columbia-0 exposed to +ultraviolet (UV) and −UV 3 weeks following onset of flowering. (a) height of inflorescence, (b) number of flowers, (c) seed biomass (d) germination rate. (Statistical analyses show differences between UV treatments ** = ≤ 0.005 ns, not significant, Student's t-test. n = 3, bar = SE mean).


Prolonged UV radiation induced severe morphological and physiological changes commensurate with those found in short-term acute-dose experiments (Kim et al. 1998; Ren et al. 2006; Hectors et al. 2007). Biochemical changes associated with protective mechanisms involving the phenylpropanoid pathway incurred costs manifested as delayed plant development, but this was followed by acclimation and recovery at 8 weeks in Col-0 with regard to leaf area and the ability to produce full-sized leaves. The rapid and continued production of flavonoids as a separate defence response incurred a continued cost in terms of plant fitness, but no effect was seen on the ability of seeds produced to germinate.

Morphological and physiological responses

Leaf trichomes offer protection via the interception, absorption and scattering of UV-B radiation (Karabourniotis et al. 1992; Karabourniotis, Kotsabassidis & Manetas 1994; Manetas 2003) and both TD and TI increased significantly in response to elevated UV-B radiation. Trichome initiation finishes before the cessation of epidermal cell division, resulting in a negative autocorrelation between TD and leaf area (Larkin et al. 1996), but the use of TI negates this. Both accessions show increases in TD and TI in response to elevated UV radiation, suggesting that UV has a direct effect on development of the leaf epidermis. Trichomes confer numerous adaptive advantages to plants, including protection against herbivory (Woodman & Fernandes 1991), changes to boundary layer dynamics, the diffusion of water vapour (Noble 1983) and the absorption of radiant energy (Ehleringer 1984). Many plant responses to UV-B radiation are thought to be triggered through co-opting defence pathways, with plants grown under elevated UV-B showing increased resistance to herbivory. There is an overlap in gene expression between UV-B and herbivory/wounding responses, both of which initiate activation of signalling mitogen-activated protein kinases (MAPKs) (Stratmann 2003). Increases in TD following herbivory have also been reported in Mimulus guttatus (Holeski 2007) and in natural populations of A. thaliana (Handley, Ekborn & Agren 2005). However, separate experiments assessing the response of TI and TD in Col-0 to simulated herbivory (wounding) indicated no change between control, and treatment (Lake, unpublished results). This lack of wounding response points to the use of a specific UV-induced signal transduction pathway and suggests a role for trichomes in protecting A. thaliana against UV radiation particularly in the early stages of leaf development when trichomes are closer together prior to cell expansion, as found in oak and olive (Karabourniotis & Bornman 1999). The effectiveness of trichomes in conferring protection against UV radiation in A. thaliana requires further investigation.

Changes in stomatal and epidermal cell frequencies also provide evidence for UV-induced disruption to the development of the epidermal layer. Decreases in SD are consistent with previously documented effects of UV-B radiation on Glycine max (Gitz et al. 2005). The systemic responses of SD to CO2 and humidity are positively related to and primarily driven by abscisic acid (ABA) concentration: a decrease ABA in mature leaves produces decreased SD in newly developed leaves (Lake & Woodward 2008). As ABA absorbs in the UV-B region and is subsequently inactivated by photolysis (Hollosy 2002), a decrease in ABA would correspond to this mechanistic relationship (Lake & Woodward 2008) A possible decrease in ABA is supported by a measured increase in gs (Fig. 3a) and subsequent decrease in WUE (measured as increased Δ13C, Fig. 3b) as would occur in the absence of ABA-induced stomatal closure. This would also appear to be a UV-B-specific-induced response as ABA absorbs in the UV-B region only (Hollosy 2002).

The reduction of SD and SI in Col-0 is accompanied by a small reduction in ED on both leaf surfaces, suggesting little disruption to cell division processes, contrary to a study using Pisum sativum (Gonzalez et al. 1998). Jasmonic acid (JA), ethylene and reactive oxygen species (ROS) are all increased in response to UV-B (Stratmann 2003). Intriguingly, ethylene and antioxidant activity were found to be involved in decreasing stomatal numbers in a systemic response to elevated CO2 (Lake, Woodward & Quick 2002).

UV-B radiation induces MAPK activity in Lycopersicon peruvianum (Holley et al. 2003). Stomatal development in Arabidopsis is regulated by two sets of MAPKs; MPK3 and MPK6 are environmentally sensitive, while upstream MAP kinase kinases, MKK4 and MKK5 are key regulators of stomatal development. A loss of function of MAPKs results in a plethora of stomata and disruption to the one-cell spacing pattern. Conversely, a gain in function results in a lack of stomatal initiation (Wang et al. 2007). While not investigated directly, it is possible that the decrease in SD and SI under UV treatment may be driven through activation of MAPKs, functioning upstream of JA (Stratmann 2003), resulting in a suppression of stomatal initiation.

SI of the fah-1 mutant increased on the abaxial leaf surface but decreased on the adaxial surface. The opposing response of the adaxial epidermal layer to UV radiation in the fah-1 mutant indicates that disruption to the phenylpropanoid pathway and the failure to synthesize specific UV-B-absorbing compounds lead to changes in the functional anatomy of A. thaliana. On exposure to UV, fah-1 had a significantly reduced leaf area and a higher ED, illustrating a lack of cell/leaf expansion but not cell division. Indeed, when UV is absent, the mutant maintains a significantly higher ED compared with the wild type. This suggests that the mutation in fah-1 (Chapple et al. 1992) disrupts normal cell expansion and that the stomatal response is specific to the adaxial leaf surface.

Derivatives of the phenylpropanoid pathway are essential intermediates in the lignification process and are responsible for physiochemical properties of the cell wall (Boerjan, Ralph & Baucher 2003; Abdulrazzak et al. 2006). A point mutation of the CYP98A3 gene of the ref8 mutant, acting upstream of both ferulic and sinapic acid production (Abdulrazzak et al. 2006), revealed that this mutant, when compared with the wild type, had a vastly reduced leaf area and a sixfold reduction in epidermal cell size. The dramatic increase in ED again confirms little disruption to cell division processes and suggests that a fully functioning phenylpropanoid pathway is a requirement for cell expansion and leaf development and agrees with the protein competition model (PCM) proposed by Jones & Hartley (1999), whereby direct competition between phenolic and protein synthesis severely affects leaf growth.


In wild-type plants, sinapic acid and its ester, sinapoyl malate (a known UV-B – absorbing compound –Landry, Chapple & Last 1995; Hemm, Ruegger & Chapple 2003), increases in response to UV. Its rapid, early accumulation under UV treatment suggests it may act as a ‘sunscreen’ absorbing/reflecting excess UV radiation, in an attempt to minimize disruption to cellular processes. This upregulation in sinapoyl malate may result in a reallocation of resources away from growth (Jones & Hartley 1999), causing the distinctly stunted appearance of 6-week-old, wild-type plants. However, Col-0 appears able to acclimate to conditions, with leaf area increasing up to and beyond that of the control plants at the 8 week stage. In 8-week-old acclimated plants, sinapic compounds are of similar abundance in both the treated and control plants, suggesting that, under control conditions, sinapic acid is a requirement of normal leaf development.

fah-1 mutants show a different metabolic response to UV radiation with an increase in ferulic acid and its ester, feruloyl malate. These mutants are unable to synthesize the enzyme ferulate-5-hydroxylase necessary to convert ferulic acid into 5-hydroxy-coniferaldehyde (Chapple et al. 1992) for production of sinapate and sinapoyl malate (Landry et al. 1995); therefore, the mutant accumulates ferulate and feruloyl malate as a result. These changes in phenylpropanoid composition result in a failure to acclimate and recover, despite accumulation of other identified flavonoid compounds, quercetin–glycoside and kaempferol–glycoside. Quercetins and kaempferols are well-documented antioxidants (Kootstra 1994) providing DNA protection against oxidative damage caused by UV-B (Kootstra 1994). Flavonoid glycosides are known to be prioritized during early development in Betula pubescens seedlings (Keski-Saari et al. 2007), with quercetin being the most efficient and kaempferol the least efficient in absorbance of UV-B (Rice-Evans, Millar & Papanga 1997). Both increased dramatically in the presence of chronic UV with kaempferol remaining at a high level throughout acclimation in Col-0, while quercetin levels were reduced. Both UV-A and UV-B are known to induce production of quercetins and kaempferols in birch leaves (Kotilainen et al. 2008); however, the reduced level of quercetin provides further evidence for acclimation as kaempferol has a lower absorbance of UV-B (Rice-Evans et al. 1997). Such concentrations under enhanced UV suggests that the role of these flavonoids is different from that of the sinapic and ferulic esters in maintaining continuous protection against oxidative damage within cells (DNA protection), rather than creating an initial barrier-type defence to shield vital processes during acclimation.

Although Hectors et al. (2007) found no evidence for upregulation of ROS scavenging gene expression when A. thaliana was subjected to 2 hourly doses per day of UV-B at 0.177 mW cm−2 (equivalent to 12.74 kJ m−2 d−1), and no induction of antioxidant enzymes was seen in pepper under UV-A (6.1 W m−2 for 27 min d−1 over 14 d – equivalent to 0.988 kJ m−2 d−1Mahdavian, Ghorbanli & Kalantari 2008), prolonged exposure to UV may induce the use of different metabolites for ROS control as a separate protective mechanism against chronic exposure. Both quercetin and kaempferol are costly in terms of carbon allocation, each containing 15 C compared to 11 C for sinapic acid and 10 C for ferulic acid. A reduction in these compounds at 6 weeks compared to 8 weeks in Col-0 again points to a re-allocation in favour of protective mechanisms over growth in the early rosette stage (Jones & Hartley 1999).

Costs of metabolic acclimation on plant fitness

Reductions in flower numbers and seed weight illustrate a cost in terms of fitness, but germination rate was not statistically significantly affected, as reported previously (Yang et al. 2004; Hectors et al. 2007). This provides evidence that the early production of phenylpropanoids and sustained production of flavonoids had an overall effect on seed production, but resource allocation in seed based protective mechanisms affords some protection for subsequent generations. In Brassica rapa, UV-B increased the concentration of UV-protective compounds in the seed testa, with a resulting decrease in testa, but not overall seed biomass (Griffen, Wilczek & Bazzaz 2004). The reduction in seed weight found here may also relate to reductions in testa biomass, as only whole seed weight was measured.


We have demonstrated major adaptations in the physiognomic and biochemical responses of A. thaliana to high levels of chronic UV exposure, that is, over the entire life cycle of the plant. These responses are generally consistent with short-term acute dose studies and lead to acclimation, which appears to be multi-phasic. Following initial retardation and acclimation due to stresses at the seedling stage, (Kim et al. 1998), we demonstrate a second phase of reduced growth in the vegetative (rosette) stage with recovery in the Col-0 wild type (i.e. to growth levels seen in non-UV treated plants), and finally, adaptation to mitigate effects on seed germination success, which together constitutes a multi-phasic response.

Metabolic profiling shows that the changes in the phenylpropanoid pathway are a key mechanism in acclimation and recovery. Initial protection occurs with the upregulation of phenylpropanoids acting as UV screening compounds. The disruption in the pathway of the fah-1 mutant and the inability to produce sinapates result in a lack of acclimation. A concomitant prolonged production of flavonoids (quercetin and kaempferol), in favour of phenylpropanoids, provides continued protection and capacity for repair. Furthermore, these flavonoids are present in the mutant, suggesting that specific protective mechanisms are offered by specific compounds.

This multi-phasic acclimation occurs at discrete stages of development and may help maintain carbon acquisition and thus recovery. The costs incurred are ultimately realized in terms of reduction of flower number and total seed biomass; however, investment in UV protective pigments maintains germination resulting in partial mitigation of the deleterious effects of UV radiation.


We gratefully acknowledge funding JAL through a Royal Society Dorothy Hodgkin Research Fellowship, BHL through a Leverhulme Trust Early Career Fellowship (ECF/2006/0492), MPD through the NERC, KJF through a University of Sheffield Studentship, Prof. T. Richardson for cross-calibration of UV sensor equipment and E. Venison and J. Wall for technical assistance.