Two lines of Medicago ciliaris L. created by three generations of spontaneous selfing in the greenhouse were used: TN11.11 and TN8.7. Seeds were obtained from the Laboratory of Legumes, Biotechnology Center of Borj-Cedria (CBBC), Tunisia.

Seeds were germinated and grown for 3 days in Agriperlite moistened with 0.1 mM CaSO₄. Three-day-old seedlings were transferred to a half strength aerated hydroponic nutrient solution for 7 days and then similar sized seedlings were selected and cultured as groups of 10 plants in 10 L of full strength aerated hydroponic nutrient solution. The composition of the nutrient solution was: 1.25 mM Ca (NO₃)2, 1.25 mM KNO₃, 0.5 mM MgSO₄, 0.25 mM KH₂PO₄ and 10 µM H₃BO₃, 1 µM MnSO₄, 0.5 µM ZnSO₄, 0.05 µM (NH₄)₆Mo₇O₂₄ and 0.4 µM CuSO₄.

Two treatments were applied: −Fe (0 mM) and +Fe (0.03 mM). Both media have a pH around 6.2. Fe was supplied as Fe (III)-EDTA. The solution was renewed every 5 days. Plants were maintained in a growth chamber with a 16/8 h day/night, 24˚C/18˚C regime and a relative humidity of 70%.

Bivalent Fe content was determined according to the method of [11] modified by [12] . After 10 days of treatment, leaves were weighted and dried at 60˚C for 72 h. Then, samples were finely crushed in a standard agate crusher in order to avoid powder contamination by Fe traces. Then, 2 g of dry samples were digested with 15 ml HCl(1N) and vigorously agitated. Digests were thereafter kept for 4 h in the vessels and then filtered before bringing them to a volume of 25 mL with distilled water. The absorbance of the solution was determined by an atomic absorption spectrophotometer (VARIAN 220FS).

Eight plants per treatment were used to determine total chlorophyll content of young leaves according to the method of [13] . One hundred mg of small discs from young leaves were incubated in 5 mL 80% acetone in darkness at 4˚C for three days (until complete chlorophyll extraction). Total chlorophyll content was determined by reading the absorbance at 649 and 665 nm.

Thylakoid membranes were isolated from leaves as described by [14] . Photosynthetic electron transport rates through the two photosystems (PSI and PSII) were measured on isolated thylakoid membranes by means of Clark-type oxygen electrode (Hansatech, UK) fitted with a circulating water jacket at 20˚C, using a light intensity of 1100 µ E m⁻²∙s⁻¹ at the surface of the water bath cell. The sample volume was 1 mL and the chlorophyll content was 10, 20 and 40 µg. The uncoupled rate of the whole chain electron transport was determined by measuring the rate of reoxidation of photoreduced methyl viologen (MV). The reaction buffer contained 100 mM sorbitol, 5 mM MgCl₂, 0.5 mM NaN₃, 3 mM NH₄Cl, 0.1 mM MV, and 50 mM HEPES-NaOH pH 8.0 [3] .

Iron-sufficient and Fe-deficient M. ciliaris leaves were adapted to dark conditions for 1h at room temperature before experiments. Chla fluorescence was measured in leaf discs using a PAM 2000 fluoremeter (Heinz Walz, Germany). The measurements were performed at room temperature (25˚C). F₀ was measured by switching on the modulated light at 0.6 kHz; PPFD was less than 0.1 µ E m⁻²∙s⁻¹ at the leaf surface. F_m was measured at 20 kHz with a 1s pulse of 6000 µ E m⁻²∙s⁻¹ of white light. The initial fluorescence (F₀) and the maximal fluorescence were determined after dark acclimation. The variable fluorescence (F_v) was taken from the formula: F_v = F_m − F₀ and the parameter used were F_v/F_m and 1/F₀ − 1/F_m [15] .

The H₂O₂ level was calorimetrically measured as described by [16] . H₂O₂ was extracted by homogenizing 50 mg leaves with 3 mL of phosphate buffer (50 mM, pH 6.8). The homogenate extract was centrifuged at 6,000g for 25 min. To determine H₂O₂ levels, 3 mL of extracted solution were mixed with 1 mL of 0.1% titanium chloride in 20% (v/v) H₂SO₄ and the mixture was then centrifuged at 6000g for 15 min. The intensity of the yellow colour of the supernatant was measured at 410 nm. Hydrogen peroxide level was calculated using the extinction coefficient 0.28 µM⁻¹∙cm⁻¹.

Volatile aldehydes were extracted in 5 ml of 2% (v/v) ethanol at 25˚C for 2 hours, under constant stirring using 0.3 g leaf fresh tissues. A 1 mL aliquot was transferred to a test tube containing 1 mL of 0.1% (w/v) 3-methyl-2- benzothiazolinone hydrazone (MBTH) before the addition of 2.5 mL of FeCl₃6H₂O solution (0.23% (w/v)). After 5 min of incubation, 6 mL of acetone were added and the absorbance was read at 635 nm [17] . The amount of volatile aldehydes produced was estimated using a formaldehyde calibration curve.

Lipid peroxidation was determined as described by [18] : 0.2 g of fresh leaves were homogenized in 4 mL of 1% (w/v) trichloroacetic acid (TCA). The homogenate was filtered through four layers of gauze and then centrifuged at 12,000 g for 15 min. One mL aliquots of the supernatants were added to 3 ml of 0.5% (w/v) thiobarbituric acid (TBA) in 20 % (w/v) (TCA) and the tubes were incubated in a shaking water bath at 95˚C for 2 h. The reaction was stopped by placing the tubes in ice. Tubes were subsequently centrifuged at 9000 g for 10 min and the absorbance of the supernatant was read at 532 and 600 nm. The concentration of the malondialdehyde (MDA)-TBA complex produced was calculated using the molar extinction coefficient of 155 mM⁻¹∙cm⁻¹.

All operations were carried out at 4˚C. Leaves were homogenised with quartz sand, liquid nitrogen and 10% (w/w) polyvinylpyrolidone (PVP) in a grinding medium (1:2.5, w:v) prepared as described below, and centrifuged at 12000 g for 30 min at 4˚C.

The extraction buffer for superoxide dismutase (SOD) and guaiacol-POD (POD) contained 220 mM Tris-HCl (pH 7.4), 250 mM sucrose, 50 mM KCl, 1 mM MgCl₂, 1% β mercaptoethanol and 0.01% (w/v) phenylmethylsulfonylfluoride (PMSF) [19] . For ascorbate peroxidase (APx) extraction, 50 mM Na-ascorbate was added to the buffer to prevent enzyme inactivation [19] . After centrifugation, the supernatant was dialyzed for 2 h against 50 mM reduced ascorbate solution (ASA) and tested for enzyme activity.

Superoxide dismutase activity was measured according to the method of [20] based on the enzyme ability to inhibit the reduction of nitrobluetetrazolium (NBT). One enzymatic unit is defined as the amount of enzyme inhibiting 50% of NBT photoreduction.

Peroxidase activity was tested using guaiacol as reducing phenolic substrate. The reaction medium contained 20 mM Na-acetate (pH 5.0), 30 mM H₂O₂, 2 mM guaiacol and an appropriate amount of enzyme extract. The rate of guaiacol oxidation was recorded at 470 nm [19] and the activity was calculated using the extinction coefficient of 26.6 mM⁻¹∙cm⁻¹.

Ascorbate peroxidase activity was determined following the decrease in absorbance at 290 nm due to the oxidation of ascorbic acid in the first 30 s from the beginning of the reaction, using the extinction coefficient of 2.8 mM⁻¹∙cm⁻¹ for ascorbate. The reaction medium contained 0.5 mM reduced ascorbate (ASA), 0.1 mM H₂O₂, 1 mM EDTA and 0.1 M HEPES-KOH buffer (pH 7.8) [21] . One enzymatic unity is equivalent to 1 µmol min⁻¹∙cm⁻¹ of ascorbic acid oxidized. To discriminate between APx and POD activity, 50 mM p-chloromercuri- benzoate (pCMB), responsible for the inactivation of APx, was added to the enzymatic reaction mixture [22] .

Non-enzymatic antioxidant activity, ascorbate and glutathione content.

Total non-enzymatic antioxidant activity was measured using the Ferric Reducing Ability of Plasma (FRAP) assay. Leaf tissue (0.5 g) was homogenized in 5 ml methanol. The homogenate was first filtered and then centrifuged at 12,800 g for 2 min. The FRAP assay was performed with FRAP reagent, i.e. 1 mM 2,4,6-tripyridyl-2- triazine (TPTZ) and 20 mM ferric chloride in 0.25 M sodium acetate (pH 3.6). An aliquot of 100 ml of leaf extract (0.5 g/10 mL in methanol) was added to 2 mL of FRAP reagent and thoroughly mixed. Next, the mixture was left at 20˚C for 5 min and the absorbance was measured at 593 nm. Calibration was obtained by means of a standard curve (25 - 1600 mM ferrous iron) using freshly prepared ammonium ferrous sulfate [23] [24] .

Extraction of ascorbate and glutathione

Leaf samples were ground in a cold mortar with liquid nitrogen and homogenized with 25 mM sulphuric acid (800 µl/400 mg fresh weight). The homogenate was centrifuged at 16,700 g for 30 min at 4˚C, and the supernatant was collected for analysis of ascorbate and glutathione.

Quantification of reduced (ASC) and oxidized (DHA) ascorbate

The assay is based on the reduction of Fe³⁺ to Fe²⁺ by ascorbic acid [25] . Fe²⁺ forms a red complex with bipyridyl that absorbs at 550 nm. DHA was reduced to ASC by pre-incubating the sample with dithiothreitol (DTT). After DTT excess was removed with N-ethylmaleimide (NEM), total ascorbate (ASC+DHA) content was measured. DHA content was then estimated from the difference between total ascorbate and the ASC. A standard curve covering the range of 2.5 - 60 µmol ASC was used.

Quantification of reduced (GSH) and oxidized (GSSG) glutathione

The contents of GSH and GSSG were determined as described by [21] based on the oxidation of GSH by DTNB (5,5'-dithio-bis-nitrobenzoic acid) to give GSSG and TNB (5-thio-2-nitrobenzene). GSSG was reduced to GSH by the action of glutathione reductase and NADPH. GSSG was assayed from the sample after the removal of GSH by 2-vinylpyridine and triethanolamine derivatizations. The rate of DTNB reduction was recorded at 412 nm for 2 min. The contents were calculated using a standard curve. The GSH content was calculated from the difference between the total glutathione and GSSG.

A two-way analysis of variance (ANOVA), with lines and treatments as factors, was performed for the whole data using the STATI-CF statistical program. Means were compared using the Newman-Keuls test at the p<0.05 level when significant differences were found. Data shown are means of five replicates for each treatment.

Medicago ciliaris plants grown under Fe deficiency conditions for 10 days developed symptoms of chlorosis. These visual manifestations were observed in both lines (TN8.7 and TN11.11) and were concomitant with a significant reduction in chlorophyll concentration (Table 1), especially for TN8.7 line (−51%). The same trend was observed for leaf Fe concentration; this decrease respect to controls was more pronounced in TN8.7 (−37%) than in TN11.11 plants (−23%) (Table 1).

Iron deficiency affected significantly the whole chain photoelectron transport rate of both genotypes TN8.7 and TN11.11 (Figure 1). As for chlorophyll and Fe²⁺ concentrations, TN8.7 exhibited the strongest reduction (57%) as compared to control plants, while for TN11.11 this decrease was less than 25%.

The measurement of chlorophyll fluorescence as the ratio F_v/F_m revealed that at the beginning of the experiment, this parameter was similar in the control leaves of both genotypes (Table 2). Under Fe deficiency conditions, this ratio decreased significantly in chlorotic leaves of both lines. This reduction was about −23% and −33% in TN11.11 and TN8.7, respectively.

For all samples, lipid peroxidation estimated by the MDA-TBA complex concentration was evaluated (Figure 2). Plants optimally supplied with Fe showed similar values. Under Fe deficiency conditions, an important increase in leaf MDA content was noted in both lines. Yet, this increase was more marked in TN8.7 (+140%) compared to TN11.11 (+95%) (Figure 2).

The concentration of volatile aldehydes as an indicator of oxidative stress degree (mg∙g⁻¹ FW) was determined (Figure 3). Our results showed that for TN11.11 no difference was found between plants supplied or not with Fe. Yet, a significant increase (+127%) was detected inTN8.7 Fe-deficient plants respect to the control.

The leaf tissue content of H₂O₂, another signal of oxidative stress, was measured. Hydrogen peroxide content significantly increased in both genotypes grown under Fe deprivation (Figure 4), suggesting the induction of oxidative stress in leaves of Fe starved plants. The highest value was detected in TN8.7 line (+68 % as compared to control plants), while for TN11.11, this increase was about 10%.

Total SOD activity was significantly stimulated under Fe deficiency in TN8.7 (+32% respect to the control), while in TN11.11 it was increased only by 13% (Table 3). To investigate whether, and to what extent, this increment was ascribable to different enzymatic isoforms, KCN and H₂O₂ were added to the enzymatic assay mixture, as different inhibitors of the three different SOD isoforms: CuZn-SOD, Mn-SOD and Fe-SOD. Results revealed that Fe deficiency induced an enhancement of CuZn-SOD (KCN-sensitive) and Mn-SOD (KCN and H₂O₂- insensitive) isoform activities, especially in TN8.7, in which the activity of both SOD isoforms was in- creased around two folds as compared to control plants. By contrast, Fe-SOD activity (H₂O₂-sensitive and

KCN-insensitive) was detected only in plants of both genotypes grown in the presence of Fe in the nutrient solution (Table 3).

Data related to POD activity demonstrated a significant decrease under Fe deficiency conditions respect to the control in both lines (−23% and −14% for TN8.7 and TN11.11 respectively). Ascorbic peroxidase exhibited the strongest decrease under Fe deficiency respect to the control in both genotypes, being −45% and −32% in TN8.7 and TN11.11 respectively (Table 4).

Leaf antioxidant activity (FRAP) of the two lines was presented in Table 5. Iron deprivation led to a significant FRAP reduction in TN8.7, (−17%) while no variation was noted for TN11.11. The results related to ascorbate (total, ASC and DHA) and glutathione (total, GSH and GSSG) concentrations in both lines under Fe-sufficient and Fe-deficient conditions are summarized in Table 5. The results revealed that TN11.11 plants subjected to Fe deprivation exhibited a significant increase in total ascorbate concentration, reaching +85% with respect to the control. The same tendency was observed for reduced ascorbate (ASC) but not for the oxidized form (DHA). Consequently, the ratio ASC/DHA showed a significant increase. However, Fe deficiency had no significant effect neither on the total glutathione concentration nor on its two forms (GSSG and GSH), resulting in a similar GSH/GSSG ratio in both control and treated plants (Table 5).

In TN8.7 line, both the total ascorbate concentration and the reduced form significantly decreased while the oxidized form did not change under Fe starvation. Nevertheless, no modification of the ASC /DHA ratio was noted under Fe deficiency stress. Similarly, the total glutathione pool and its reduced form decreased in Fe- deficient leaves (respectively −19% and −25% respect to the control), whereas the oxidized form (GSSG) was similar to the control. Also in this case, the GSH/GSSH ratio remained constant (Table 5).