Changes in the environment may represent so-called 'stress situations' in plants. 'Stress' is typically characterised by excessive formation and release of reactive oxygen species (ROS; e.g. superoxide anion O₂^-·, hydrogen peroxide H₂O₂, singulet oxygen O^·, and hydroxyl radical OH^·) and consequently causes changes in the redox environment on cellular level 1. These phenomena are summarised with the term 'oxidative burst' 2. An oxidative burst is followed by multiple responses in transcription 34, translation 56, protein activity 78, metabolism and possibly programmed cell death 910.

ROS formation always occurs during normal growth and development, particularly in sub-cellular locations with high enzymatic redox turnover. Thus, the formation and destruction of ROS are well balanced in plant cells and under the control of a complex antioxidative system 1112. The antioxidative system mainly consists of antioxidative enzymes (e.g. APX, GPX, SOD, CAT, GR) catalysing electron transfer to ROS using low molecular weight antioxidants (e.g. ascorbate, tocopherol, GSH) as electron and proton donors 1314.

However, during an 'oxidative burst' this equilibrium becomes unbalanced and the organism is forced to adjust its antioxidative system in order to cope better with the current or a future stress situation. A pre-requisite for this is a signalling network that is able to switch antioxidative and metabolic redox capacities. Consequently, ROS are also considered to be signalling components which specifically trigger antioxidative responses 1516171819. Many studies show how and where the formation of ROS occurs (e.g. for review see 20). Although particular effects on the antioxidative system are well characterised, little is known about general metabolic effects.

Transcript profiling and proteomic approaches are used for tracking adaptations in antioxidative system in response to environmental changes. However, for proof of functional changes a third trace of evidence using metabolic profiling has to be adopted 212223. Here, we present a first approach to quantify changes in the antioxidative system of plants after abiotic stress challenge using 'summary parameters'. To achieve this, we have optimised widely used chemiluminescence assays.

These assays allow the quantification of:

1) the total antioxidative capacity (TAC) of low molecular weight metabolites,

2) the luminol converting peroxidase activity (LUPO) in plant tissue,

3) the total superoxide scavenging activity (SOSA) of high molecular weight compounds including SODs.

All assays yield information about the general antioxidative status and about specific aspects of the antioxidative system, rather than exact data for single antioxidant species. For each luminometric assay information is given on how to tune the sensitivity and how to fit assay performance to the requirements of the biological specimen under study.

We compare results obtained from probing Lepidium sativum under different abiotic stress situations and show that common photometric assays for catalase (CAT) and glutathione reductase (GR) activities can easily be run in parallel.

Antioxidative power is generally defined as the ability to scavenge ROS and antioxidative capacity is expressed in terms of concentration of a pure antioxidative substance. TAC can thus be used as a marker to detect changes of antioxidative metabolism during oxidative stress.

The general basis of TAC assays is a redox reaction driven by a ROS (typically H₂O₂). The signal output produced, is subsequently quenched by the addition of a sample with ROS-scavenging properties. Luminol (L = 3-Aminophthalhydrazide) is a frequently used reagent which emits light when oxidised by H₂O₂ to aminophtalate (AP) under alkaline conditions (see Fig. 1.1 in additional file 1). This chemiluminescence reaction can be catalysed by peroxidases (POs) such as horseradish peroxidase (HRP). HRP is a widely used enzyme that oxidises phenolic compounds with hydrogen peroxide (H₂O₂) as oxidant 24. The HRP-catalysed luminol reaction involves three steps by which the HRP protein undergoes conformational changes until functional HRP is regenerated (Figure 1, 2526).

Molecular oxygen (O₂) is not involved in this reaction 2728 as long as there is a reactant 29. However, other additives are required for the HRP reaction that either serve as primary electron donors or as excipients that enhance enzyme turnover and stability. Calcium is an essential cofactor for HRP 2530. Phenolic electron donators with higher affinity to HRP than luminol (e.g. iodophenol) enhance the luminescence drastically 31. Also a surfactant is beneficial as it stabilises the protein, and enhances the light output (3233, Fig. 1.2 in additional file 1).

The pH optimum of the reaction is around pH 8.7 (Fig. 1.3 in additional file 1). Although the light reaction is driven by H₂O₂, an excess of it inactivates HRP (2633, Fig. 1.4 in additional file 1).

The reaction is started by the addition of H₂O₂ (details in Methods section). When a constant light output is recorded, an antioxidant containing sample is added (Figure 2). The luminol-HRP electron transfer is then inhibited due to the competition of the antioxidants with luminol.

Light output is quenched according to the amount of antioxidant. When the antioxidative capacity (AC) of the added sample is exhausted, the HRP-catalysed electron transfer from luminol continues and light output recovers. The amount of quenching (see Eq.1 additional file 1) and the time required for signal recovery after sample addition (i.e. the time point with maximal signal increase after oxidation of the sample, Figure 2B) are dependent on the amount of antioxidant added. The recovery times of specific amounts of a pure antioxidant are used to calibrate the assay (see Fig. 1.5 in additional file 1).

For each sample to be quantified the recovery time is measured and the result is given in terms of molar concentration of the pure antioxidant used for calibration. The result has to be related to fresh weight, dry weight, or total protein content of the sample.

HRP is employed in a wide range of analytical assays 34. In particular, the HRP-based TAC assay is used in medical research and food science, mainly to probe the TAC of blood serum 35, beverages 36, vegetables, and other foods 37.

Tissue samples from plant material always contain the full complement of their antioxidative factors. This includes peroxidases and catalases which may compete with HRP in the assay and thus influence the light output and lead to erroneous TAC values. Hence, the low molecular weight antioxidants to be quantified have to be separated from enzymatic activities. Therefore membrane filtration with MWCO of 10 kDa has been found to be useful.

Peroxidases are widely distributed among living organisms and have many physiological functions 1438. They exhibit a broad substrate specificity and require organic substrates for catalysis 39. Many other peroxidases besides HRP also accept luminol as electron donor (L, see Fig. 2.1 in additional file 2). Luminol converting peroxidases (LUPO) can readily be quantified by their light yield (Fig. 2.2, 2.3, 2.4 in additional file 2). Since low molecular weight H₂O₂-scavengers may interfere with the LUPO reaction, the sample has to be dialysed before use (details in Methods section). The recorded light output is linearly correlated to the amount of peroxidase and, if required, can be expressed in equivalents of a standard peroxidase (Fig. 2.4A in additional file 2).

The superoxide anion (O₂^-·) is formed by single electron transfer from over-reduced redox enzymes to molecular oxygen. It has a short lifetime in living cells 40 and is disproportionated to H₂O₂ and molecular oxygen. Plants possess several superoxide dismutases (SODs) scavenging superoxide anions enzymatically 414243. Additionally, non-enzymatic O₂^-·-scavengers are also found in plants 44.

Like the TAC assay, the method for assaying the superoxide anion scavenging activity (SOSA) presented here, is based on the quenching of chemiluminescence. Here, the light-yielding substrate is coelenterazine (CTZ), a specific O₂^-·-indicator (4546, Fig. 3.1 in additional file 3).

Due to its very short lifetime, O₂^-· has to be generated in situ. The most convenient method to generate O₂^-· employs xanthine oxidase (XOD) 47. While XOD is generating O₂^-· (Fig. 3.1A in additional file 3) and thus producing light in presence of CTZ (Fig. 3.1B in additional file 3), any O₂^-·-scavenger in the assay reduces the steady state O₂^-· concentration and thus light output. This quenching of CTZ luminescence is used to quantify the scavenger activity. In the dialysed sample non-enzymatic high molecular weight scavengers 48 can be distinguished from enzymatic (i.e. SOD) scavenging activities (Figure 3) (also see Fig. 3.2 in additional file 3).

For a SOSA assay, several factors have to be considered:

1) The XOD needs a physiological pH and calcium buffering for high superoxide anion yield. From all available XODs the one from bovine milk does produce sufficient O₂^-· 49.

2) There are many CTZ analogues available for use as O₂^-· indicators (Tab. 1 in additional file 3). Each produces different luminescence yield (Fig. 3.3A, B in additional file 3). However, the important feature for the assay is the signal-to-background luminescence ratio (50, Fig. 3.3C in additional file 3). CTZ needs careful handling since it is easily oxidised by ambient oxygen when in solution (details provided in the Methods section and in four additional files). The CTZ concentration determines the duration of constant light output (Fig. 3.4 additional file 3).

3) A surfactant is needed to maximize luminescence yield.

Adaptation to cold and formation of freezing tolerance involves a cascade of cellular events beginning with Ca²⁺- and IP₃-mediated signal transduction, transcription factor activation and gene induction 51525354, leading to metabolic changes. Cold-induced metabolic changes have been studied and many metabolites and enzymes involved in cold adaptation have been identified 21545556. However, the whole activation cascade is expected to work very slowly at 0°C. Thus, it seems that other posttranslational events are involved. This has been shown in particular for GR. Here 57, an increase in substrate affinity after cold treatment is reported rather than an increase in enzyme level or in GR transcripts.

We used five day old garden cress (Lepidium sativum) seedlings and applied a cold period (0°C) over night (12 h) and allowed recovery (12 h) under normal growth conditions thereafter.

During a 12 h cold period, TAC is increased (Fig. 5A) and enzymatic activities (particularly peroxidases) are activated (Figure 5B). This reverts back to normal level during recovery. In contrast, the superoxide anion scavenging activity (SOSA) appears to be unaltered (Figure 5D). The enzyme activities (LUPO, GR and CAT) are also up-regulated during the cold period (Figures 5B, C, E). Most prominent is LUPO activity which increases by about 80% compared to control plants. CAT up-regulation seems to continue during the recovery period, while GR is regulated back to normal and LUPO is even down-regulated below normal. These results are in line with previous reports 5859 which also showed induction of CAT and PO on the metabolic and transcriptional level during the first 12 h of cold.

When comparing the polygons in Figure 4, it becomes obvious that cold treatment (Figure 4A) makes a higher impact than all other stress factors (Figures 4B–D). This has already been noticed by others 21. For many physiological responses the rate of temperature change (i.e. cooling rate) and not the steady state temperature is of importance (60, for review see 61). This has in particular been shown for Ca²⁺ signalling 5262. Thus, it would be no surprise if this was also the case for the antioxidative parameters studied here. The substantial increase in the antioxidants and peroxidases during cold treatment (Figures 5A, B) is probably due to the cold shock applied. A more moderate treatment (slow cooling) may have less impact on the antioxidative system.

Heat directly leads to oxidative damage. Cellular protection mechanisms against this involve different molecular cues. Up-regulation of genes and expression of heat-shock proteins (HSP) are well established responses to heat stress in plants 6364.

For heat treatment we challenged Lepidium sativum seedlings for 6 h with 42°C. We chose a dark incubator to avoid light-induced additive effects which have been reported earlier 64. The treatment was started in the morning and the plants were harvested about midday. For recovery a period of 6 h under normal growth conditions was allowed.

The fingerprint (Figure 4B) reveals a mobilisation of TAC, SOSA, CAT, and GR. Nevertheless, there is no significant change in the area of the black polygon caused by massive reduction of LUPO (Figure 6B). The first prominent finding, up-regulation of SOSA (Figure 6D), implies that there is apparently a demand for O₂^- scavengers, suggesting excessive formation of superoxide during heat treatment. Thus, in order to cope with high concentrations of hydrogen peroxide generated during superoxide dismutation, the plant seems to mobilise low molecular weight antioxidants and catalases as well (Figures 6A, E). The second finding is a massive down-regulation of LUPO (Figure 6B). It returns to normal level when the plant is allowed to recover from heat treatment. In accordance with our findings, are results obtained by heat treatment of Arabidopsis plants. Panchuk et.al. 65 found, that the total activity of ascorbate peroxidase in Arabidopsis was strongly reduced after several hours exposure to 44°C. Although peroxidases are generally believed to be heat stable, it has been shown that they suffer even from moderately raised temperature (42°C) 66. In order to prove that this is also the case for the LUPO activity in Lepidium we performed a control experiment (Fig. 2.5 in additional file 2). The results show that LUPO is heat sensitive. Thus the reduction in LUPO (Figure 6B) is probably due to direct heat inactivation.

Different mechanisms are involved when plants respond to salt stress, namely defence against the osmotic effect of high salinity and against the ion toxicity. Apart from proline 6768 no specific profiling of antioxidative metabolites has to our knowledge been reported for salt stress.

For salt stress treatment in this study, hydroponically grown Lepidium sativum seedlings were exposed to high salt concentrations (150 mM NaCl) for 24 hours. Afterwards a 24 h recovery period in 0.5 × MS medium was applied.

There are numerous studies where salt stress induced responses of antioxidative enzymes were investigated. On the one hand, our results backup previous findings 697071, since they show that SOSA is up-regulated during salt stress (Figure 4D). On the other hand, our results contrast with other studies. Peroxidase activity, for instance, increased significantly under salt stress in Trigonella species 72. This inconsistency (Figure 7B) can be explained by the different conditions and time scales used for challenging plants with abiotic stress. Other previous work revealed that plants having a higher capability to neutralize ROS also have a higher salt tolerance 73. In particular, Demiral and Türkan 74 demonstrated that a salt-tolerant rice cultivar (Pokkali) up-regulated CAT and APX activities during stress, whereas the salt-sensitive IR-28 does not exhibit this response. Gosset et al. 75 reported a significant increase in the activity of antioxidative enzymes in the salt-tolerant cultivar of cotton during salt stress. These results also indicate that fingerprinting the antioxidative system may allow the distinction of tolerant from non-tolerant lines. Here, salt stress does not influence TAC (Figure 7A), but inhibits GR and CAT (Figures 7C, E).

After recovery the salt stress response fingerprint does not return to normal (Figure 4D, grey polygon), as seen with all other abiotic treatments (Figures 4A–C). This is because the 'recovery treatment' (exchange of salt solution by normal nutrition medium) implies an additional challenge, (hypoosmotic stress), which provokes another shift of the antioxidative system. TAC is restored and even significantly up-regulated (Figure 7A and Figure 4D). Here, it cannot be distinguished whether this is an effect of the concurrent hypoosmotic shock or simply a relief of the 'high ion' effect on the involved enzymes.

Drought stress (i.e. water deficiency) generally depends on osmotic pressure changes. Since 'drought' cannot be precisely adjusted like temperature or salt concentration, in many previous studies osmotic stress produced by mannitol solutions was taken as a surrogate for drought 76.

Here, we avoided mannitol treatment since it is known to be a ROS scavenger 77, which would interfere with any antioxidative fingerprint experiments. We therefore applied a 24 h drought period by total withdrawal of nutrient medium from hydroponically grown Lepidium seedlings. Withered plants completely regained turgor during the following recovery period.

Gogorcena et al. 78 demonstrated that in pea (Pisum sativum) nodules drought causes a decrease in all relevant antioxidative enzymes and markers. They conclude that the decline of antioxidative capacity is due to an exhausted NAD(P)H pool. In contrast, Moran et al. 79 and Zhang and Kirkham 8 both reported an up-regulation of peroxidases and down-regulation of CAT in response to drought. The former is in line, the latter opposite to our findings (Figure 8B, E). However, the results are hardly comparable since their experiments 79 were performed on a different time scale (days) compared with those presented here (hours). As with heat treatment (Figure 6F) drought induced different effects on the various antioxidative parameters (Figure 8F) so that the area of the corresponding polygon remains almost unchanged (Figure 4C). The increase in SOSA (Figure 8D and Figure 4C) is also seen with heat and salt treatment (Figures 6D, 4B and Figures 7D, 4D). All three stressors produce an increase in cellular ionic strength that may increase a demand for superoxide scavengers. The reduction in LUPO (Figures 8B, F and Figure 4C) is in contrast to 79 and 8, suggesting either a loss of function at high ionic strength or a down-regulation due to a decreased demand of peroxidase activity in Lepidium. To decide this, we performed the LUPO assay in buffers of different ionic strengths (5 to 500 mM NaCl) and found no loss of function at high salt concentrations (data not shown). Therefore, a demand-driven down-regulation is more likely. This conclusion is also valid for salt stress (Figures 4D, 7B). The slight expansion of the polygon after recovery (Figure 4C, grey polygon) suggests a formation or mobilisation of over-all antioxidative power. This may constitute a memory for future challenges.

The current understanding of metabolic adaptation in response to abiotic challenges assumes genetic induction and the formation and mobilisation of antioxidative power. From the work presented here, another facet of this response emerges, namely a shift of function, i.e. the down-regulation of a selected function in favour of another.

If abiotic stress challenge only led to the formation of antioxidative power, then this would appear as an increase in area when plotting the changes in any set of screened parameters as a polygon. This is true only for cold stress treatment (Figure 4A). For all other abiotic stimuli no remarkable increases in area are seen (Figures 4B–D). Here, any expanding corner representing an up-regulated parameter is compensated by a contraction elsewhere in the polygon representing a down-regulation. This is particularly obvious when comparing LUPO and SOSA. These two parameters run anti-parallel under all investigated stress conditions, i.e. SOSA is increased while LUPO is down-regulated and vice versa. This is reasonable, since SODs form H₂O₂ and peroxidases may also produce superoxide 39.

This also suggests that the plant is limited in mobilising resources under stress and mainly responds to abiotic stress by shifting antioxidative functions, i.e. up-regulation of necessary scavengers and a down-regulation of dispensable functions. Consequently, the fingerprint may yield information as to what kind of ROS is mainly involved in each type of abiotic stress. Thus, if there is no increase in CAT, then there is probably no demand for additional H₂O₂ scavengers. Conversely, if there is a shift towards SOSA then superoxide anions may be the major ROS.

Here, we show that each kind of abiotic stress gives rise to a specific antioxidative defence (Figure 4). However, redundancies within the group of screened parameters reduce the informative value. Therefore it is critical to identify these potential redundancies.

Among the five parameters presented here (i.e. TAC, LUPO, SOSA, CAT, GR) redundancies appear to be present: TAC and GR always run in parallel. This coincidence suggests that the majority of low molecular weight antioxidants (i.e. TAC) in the plant is represented by GSH. Further, if an interdependency of LUPO and SOSA can be consolidated then their anti-parallel behaviour could also be regarded as redundant.

On the one hand low redundancy is helpful within a multi-parametric set of assays to increase the informative value. On the other hand, however, some redundancy can be used as an internal cross-check to identify errors and to minimise misinterpretation.

Differences in controls (i.e. plants not challenged with abiotic stress, shown by white bars in Figure 5, 6, 7, 8) can be attributed to developmental and circadian variations. For instance, controls running in parallel with the heat stress experiments were harvested at midday and in the evening (6 h interval). They show considerable differences in all screening parameters. In particular, LUPO is up-regulated at midday, when maximum photosynthesis (ROS production) is reached, and down-regulated in the evening. Differences are less obvious in the controls of the salt and drought stress experiments (Figures 7, 8 white bars). Here, all samples were harvested at the same time of the day. In contrast, the controls of the cold stress experiments have been harvested in 12 h intervals (evening and morning). Here, GR activity in the 'morning sample' (Figure 5C, last column) is slightly higher than in the 'evening sample' (Figure 5C, second column). This could be an indication for circadian variations of GR activity, as has been discussed previously 80.

Antioxidative compounds can be very different among plant species 81. Hence, it is more meaningful to monitor a global parameter such as TAC before going into a detailed analysis. Nevertheless, many other plant metabolites with specific antioxidative properties can also be studied separately. However, the analysis of each component requires specific processing of plant material and analytical methods. A gain in information can justify expenses and effort, but, as mentioned above, an increase in specificity would also generate more redundant data.

Apart from salt stress (Figure 4D, discussed above) the areas of all grey polygons in Figure 4, representing the antioxidative status after recovery, are close to the red pentagons (i.e. zero alterations). This shows that the antioxidative system is very flexible and that any changes are quickly adjusted to normal when stress factors are removed. Nevertheless, heat (Figure 4B) and drought (Figure 4C) show a slightly increased polygon area even after recovery. This could be interpreted as a gain in tolerance and is in line with findings that abiotic stress stimuli leave an imprint on plant metabolism 7682.

Robustness of a plant is defined as the ability not only to withstand adverse conditions but also to propagate, and to produce a yield of sufficient quality in non-optimal environments. Tolerance to abiotic stress and robustness have often been directly linked to the efficiency of the antioxidative system (83 and references cited therein; 84). A more robust plant is therefore able to activate its antioxidative systems faster, or to shift antioxidative functions more effectively.

There have been many attempts to breed robust, stress tolerant crop plants, even by transgenic approaches (http://www.plantstress.com; 858687). The success of this endeavour has to be verified by long lasting and costly growth-yield experiments. However, if a certain pattern or fingerprint of the antioxidative system, as depicted here (Figure 4), could be shown to strictly correlate with robustness, then screening would become easier. In plant breeding projects high throughput screening is needed to distinguish stress tolerant from sensitive breeds at an early stage of growth 88. Thus, metabolic 'summary parameters' may be implemented as quantitative traits when screening for robustness among different ecotypes, novel transgenics or breeds of a plant species.

Directed genetic interference with the antioxidative system sometimes leads to improvement of stress resistance 89. However, unexpected results are often obtained 90919293, requiring detailed analysis of what has changed in the overall metabolism. In such cases, a fingerprinting approach could also help to find explanations.

Bradford Assay Reagent (BioRad #500-0069)

Calcium chloride (CaCl₂ 6 H₂O; Riedel deHaen #12074; MW = 220 g/mol)

Coelenterazine (NanoLight Technologies #303; MW = 423 g/Mol)

Di-potassium hydrogen phosphate (Roth #T875; MW = 174 g/Mol)

EGTA (Sigma #E4378; MW = 340 g/Mol)

Ethanol abs. (Roth # 9065)

Glutathione – oxidised (GSSG; Roth #6378; MW = 612 g/Mol)

Horseradish peroxidase (HRP; Sigma #P6140, ca. 2 kU/ml)

Hydrochloric acid (HCl; Roth #4625; 1 M, i.e. 34% 1:10 diluted in H₂O)

Hydrogen peroxide solution (H₂O₂; Merck # 1.08597; 30% = 8.8 M; MW = 34 g/Mol)

Hypoxanthine (Fluka #56700; MW = 136 g/Mol)

Iodophenol (Fluka #58020, MW = 220 g/Mol)

Luminol (Fluka #09253; MW = 177 g/Mol)

Murashige and Skoog (M&S) medium (Duchefa, #M0222)

NADPH tetra-sodium salt (Roth #AE14; MW = 833 g/Mol)

Potassium di-hydrogen phosphate (Merck #1.04873; MW = 136 g/Mol)

Potassium hydroxide (KOH; Roth #6751; MW = 56 g/Mol; 5 M in H₂O)

Sodium chloride (NaCl; Fluka #71378; MW = 58 g/Mol)

TRIS ultra pure; (ICN Biomedicals #77861; MW = 121 g/Mol)

Triton^® X-100 (Sigma #X100; MW = 647, Liquid ρ = 10.7 kg/L ≅ 1.65 M)

Trolox (Aldrich #23,881; MW = 250 g/Mol)

Xanthine oxidase (from bovine milk; Sigma #X4500)

Seeds from Lepidium sativum were obtained from Sperli (#42.953, Germany, http://www.sperli-samen.de). For surface sterilization 4 gram of seeds were vortexed 2 min. in 80% ethanol and dried on filter paper.

For cold and heat treatment seeds were sown on 2 layers of synthetic capillary matting (medical fleece "Rolta^®-soft" # 932048, Hartmann, Germany, http://de.hartmann.info/) soaked with 0.5 × Murashige and Skoog medium (Duchefa #M0222). Seedlings were grown in a mini propagator (26 × 11 × 7 cm, Windhager^®; Austria; http://www.windhager.at) at 21°C with a 16 h photoperiod (50 μE, white fluorescent lamps Osram L36 W/77) for five days before treatment with abiotic stress.

For salt and drought treatment plants were hydroponically grown on 0.5 × M&S medium in sterile Growtek™ culture vessels (#BEL1768; BLD Science^®, NC, USA; http://www.bestlabdeals.com) as above. Hydroponical growth enables easy removal and exchange of nutrient medium.

Cold and heat stress were applied by placing the mini-propagator for 12 h at 0°C or 6 h at 42°C, respectively, in the dark. Recovery times were allowed for 12 h and 6 h, respectively, in the growth room. Whole plants were harvested and processed.

Salt stress was induced by exchange of the nutrient medium with 150 mM NaCl in 0.5 × MS medium. Salinity was applied for 24 h. For recovery, the salt solution was removed. Roots were rinsed with tap water before plants were allowed to recover for 24 h in 0.5 × MS medium and under conditions as before.

For drought stress, the nutrient medium was completely withdrawn for 24 h. This period is sufficient to produce an appreciable limp appearance and yet short enough to maintain ability for full recovery. For recovery (24 h) 0.5 × MS medium was replaced. From plants grown in Growtek™ vessels only the green parts were harvested and processed.

Plant material was harvested after stress treatment and after recovery time. Untreated controls were run in parallel (experimental design see Fig. 4.1 in additional file 4). Portions of 4 grams fresh weight were snap-frozen in liquid nitrogen and stored at -80°C until processed.

During further processing (see Fig. 4.2 in additional file 4) contact of samples with ambient oxygen was minimised. Frozen plant material was ground in liquid nitrogen with a mortar and pestle. Ten volumes ice-cold degassed buffer (100 mM TRIS/HCl pH 8.6 + 2 mM CaCl₂ + 1 mM Triton^® X-100) was added to the powder (ml/g), vortexed for 2 min, and filtered through a fluted paper filter in a 0°C cabinet under N₂-atmosphere. This crude extract was aliquotted for membrane filtration (TAC) and for dialysis (LUPO, SOSA, enzyme assays).

For analysis of low molecular weight total antioxidative capacity (TAC) the crude extract was passed through a membrane filter (MWCO = 10 kDa, VivaSpin 20, VivaScience, Germany, http://www.vwr.de) by centrifugation at 0°C.

Since low molecular weight antioxidants and phenolics can interfere with enzymatic assays of LUPO, SOSA, CAT, and GR 94, these compounds were removed by dialysing the crude extract in dialysis membrane tubing (MWCO = 10 kDa; Roth #E668.1) for 30 min per 100 μl dialysate against twohundred-fold volume of ice-cold assay buffer.

Assay results have to be displayed with reference to an invariable parameter. Therefore, in many previous studies plant fresh weight or dry weight was used. However, dry weight is not invariable with salt treatment due to considerable ion uptake and fresh weight may vary with drought treatment due to loss of water. Consequently, we employed the total protein content as a reference parameter, knowing that this may also vary within certain limits. Thus, after dialysis the protein concentration was determined by the Bradford assay 95. Before assaying, all samples were supplemented with buffer in order to equalise differences in protein concentrations.

Luminescence assays were performed with a simple chemiluminometer (PMT 9829A, Electron Tubes Ltd. Ruislip, UK) equipped with a light tight sample housing to hold vials in front of the detector 96.

The assay buffer was 1 mM CaCl₂ + 100 mM TRIS/HCl pH 8.6. Iodophenol was dissolved in ethanol to give a 100 mM colourless stock solution. Luminol dissolved in 5 M KOH gave a 1 M stock solution. The assay mixture (sufficient for ca. 2000 samples) was prepared by adding 20 μl ethanolic iodophenol stock, 500 μl luminol stock, 100 μl HRP suspension, and 660 μl Triton^® X-100 to 1 L of assay buffer. This mixture can be used diluted to increase assay sensitivity and thereby meet samples with lower TAC (Fig. 1.6 in additional file 1). It is stable for several weeks, when stored at 4°C (Fig. 1.7 in additional file 1). H₂O₂ (1.1 mM) solution was prepared by diluting H₂O₂-stock (30%) 1:8000 in 100 ml of assay buffer. The light emitting reaction was started by mixing 0.5 ml H₂O₂ with 0.5 ml of assay mixture. After establishment of a constant signal, the luminescence was quenched by injecting the sample (0.5 ml). Light recording was continued until luminescence recovery (details in additional file 1).

An assay buffer of 100 mM TRIS/HCl pH 8.6 + 2 mM CaCl₂ + 1 mM Triton^® X-100 was used. Luminol (1 mM) solution was prepared by diluting 1 M alkaline luminol stock solution in assay buffer. A 17.6 mM H₂O₂ solution was prepared by diluting H₂O₂ (30%) 1:500 in assay buffer. Dialysed samples were diluted 1:100 in assay buffer. 0.5 ml of diluted sample was mixed with 0.5 ml luminol solution and background luminescence was recorded. The light reaction was started by adding 0.5 ml 17.6 mM H₂O₂ solution. Counts per second (cps) were recorded for several minutes and light output integrated (details in additional file 2).

An assay buffer containing 100 mM potassium phosphate pH 7.4 + 0.1 mM EDTA + 6 mM Triton^® X-100 was used. CTZ (50 μM) solution was prepared by diluting a 5 mM methanolic stock solution in thoroughly de-gassed potassium phosphate buffer (100 mM pH 7.4). CTZ solution was aliquotted to 0.25 ml portions in luminometer vials under an oxygen free atmosphere and capped. Hypoxanthine (1 M) stock solution was prepared in 5 M KOH. HX (1 mM) was prepared by dilution this alkaline stock 1:1000 in assay buffer. Xanthine oxidase mix was prepared by diluting XOD suspension 1:3000 in assay buffer. Dialysed samples were diluted 1:8 in assay buffer. Dark-background from uncapped CTZ aliquots was recorded before injecting 0.25 ml HX-solution. Background luminescence was recorded for a while and then 0.5 ml XOD-mix was injected to initiate the superoxide-CTZ reaction. 0.5 ml of diluted dialysed sample was injected to quench luminescence and light recording was continued for several minutes. Before assaying for SOSA the dialysate was checked for TAC to ensure that no low molecular weight antioxidants adulterate the assay.

The assay buffer consisted of 100 mM potassium phosphate pH 7.4 + 8.8 mM H₂O₂. The CAT-assay was based on the UV-absorption of H₂O₂ 97. The consumption of H₂O₂ after injecting the CAT-containing sample was recorded at λ_ABS = 240 nm in a spectrophotometer (Ultrospec 2100pro, GE-Healthcare Europe, München, Germany). The linear decay in H₂O₂ was calculated (Figures 5E, 6E, 7E, 8E).

An assay buffer containing 100 mM potassium phosphate pH 7.4 + 0.1 mM NADPH + 1 mM GSSG was used. The GR-assay is based on the absorption of light by NADPH, the co-substrate of the GR 98. The fall in absorption of λ_ABS = 340 nm was recorded with a spectrophotometer (as above) after injecting the GR containing sample. The linear decay of NADPH signal was calculated (Figures 5C, 6C, 7C, 8C).