Cell suspension cultures of Arabidopsis thaliana L., Heynh., ecotype Landsberg were maintained in MS (Murashige and Skoog, 1962) medium supplemented with 30 g l^-1 sucrose, 0.5 mg l^-1 NAA (naphthaleneacetic acid), 0.05 mg l^-1 Kinetin. Cell suspensions were subcultured in 500 ml flasks at 15-day intervals by inoculating 2 ml of packed cell volume in 50 ml of fresh medium.

MS medium (3.150 l) containing 30 g l^-1 sucrose, 0.5 mg l^-1 NAA, 0.05 mg l^-1 Kinetin was inoculated with 350 ml of 14-day-old shake flask suspension in a 5.0 l stirred bioreactor (BioFlo 110, New Brunswick Scientific, Edison, NJ, USA). Cultivation was performed at 25°C, pH 5.6 and 80 rpm agitation speed, under continuous fluorescent white light (50 μmol photons m^-2 s^-1). Before the anoxic stress, cells were cultivated under 20% dissolved oxygen (DO) of air saturation, automatically obtained by the gas mix controller (New Brunswick Scientific, Edison, NJ, USA).

Arabidopsis cells grown in a stirred bioreactor for 8 days under 20% DO were subjected to anoxia for 4 h, by stopping aeration and fluxing with nitrogen into the vessel (0.01% DO). During the whole period of treatment, the bioreactor was maintained in the dark. Thereafter they were re-oxygenated by restoring the previous aeration conditions for 20 h. All the analyses were performed on samples taken at different times: T0, normoxia; T1 and T2, anoxia for 2 and 4 h, respectively; T3 and T4, 2 and 20 h after re-oxygenation, respectively. T1 and T2 samples were collected under nitrogen flux and immediately used for avoiding re-oxygenation.

Arabidopsis cell suspension cultures were stained with the Evans Blue dye and cell death was determined by spectrophotometric analysis according to Carimi et al. (2003). For each time, three independent experiments were performed with each assay done in triplicate. Dead cells were also analyzed by light microscopy according to de Pinto et al. (1999). For each treatment, 500 cells were examined.

For ASC and glutathione determination, 0.3 g cells were homogenized with cold 5% metaphosphoric acid at 4°C at 1:3 ratio (w/v) in order to obtain a deproteinized extracts. After centrifugation at 20000 g for 20 min, the supernatants were collected and used for ASC and GSH analysis according to Paradiso et al. (2006).

Extraction and analysis of tocopherols were carried out as previously described (Caretto et al., 2002). Briefly, the method consisted of an alkaline hydrolysis (potassium hydroxide 60%) followed by extraction with n-hexane-ethyl acetate (9:1). Chromatography separation was performed by using a Beckman HPLC Analytical System. A RPC18 Beckman Ultrasphere column was used with methanol (98%) as the mobile phase. Two programmable detectors, an ultraviolet-visible spectrophotometer (set at λ: 290 nm) and a spectrofluorimeter (λ excitation: 289 nm; λ emission: 325 nm) were connected in series to determine tocopherols. The tocopherol content was calculated by means of standard calibration curves. Each experiment was carried out with at least three replicates.

A confocal laser scanning microscope system, CLSM (Zeiss LSM Pascal, Carl Zeiss Inc., Germany), equipped with He-Ne and Ar lasers and coupled to Axiovert 200 inverted microscope (Zeiss, Germany) was used to detect ROS, NO, and H₂O₂ in anoxia-treated and untreated Arabidopsis cells by using specific fluorescent probes.

Reactive oxygen species were detected by using 25 μM carboxy-H₂DCFDA in DMSO, from Image-iT^TM LIVE Green ROS Detection Kit (Molecular Probes, Invitrogen, Ltd., Paisley PA4 9RF, UK) following the manufacturer instructions. Two hundred microliters of anoxia-treated cell cultures, at different times, and the relative controls, were incubated for 30 min at 25°C, in darkness, with 25 μM carboxy-H₂DCFDA probe. Then cell suspensions were centrifuged at 2000 × g, washed twice with 1 ml of fresh medium and finely suspended in 200 μl of medium. Cells were immediately observed by CLSM.

Hydrogen peroxide (H₂O₂) presence was detected by DHR probe (Life Technologies, Carlsbad, CA, USA). Two hundred and fifty microliters of anoxia-treated and control cells were incubated 10 min with 1 μl of DHR, in the darkness. The medium was then removed and the cells were washed twice with fresh medium and immediately imaged by CLSM. The specificity of the staining for H₂O₂ was tested in a parallel experiment by adding 30 U ml^-1 of CAT before the addition of DHR.

For NO detection a 250 μl of cell culture sample, were incubated for 10 min at 25°C, in darkness, with 1 μl of DAF-2 DA (Calbiochem-Novabiochem) diluted in medium, then cells were centrifuged at 2000 × g, washed twice in fresh medium and immediately imaged by CLSM.

Carboxy-H₂DCFDA, DAF-2 DA, and DHR probes were excited with the 488 nm line, and filtered through a 505–530 nm band pass filter, fluorescence was displayed as green false color. The second channel collecting emission beyond 650 nm by a long band-pass filter (excited at 488 nm laser line) was used to collect the autofluorescence of chlorophylls and the highly excited R123, and was displayed as red false-color. Background staining, routinely negligible, was controlled with unstained cells, which showed only the red autofluorescence of chloroplasts (data not shown). Confocal images were recorded using Plan-Neofluar 20×/0.5 and Plan-Neofluar 40×/0.75 objectives. A minimum of nine randomly taken fields, per treated and not treated cell cultures, was recorded. LSM Pascal 5 software was used to record and match confocal images.

Cells were ground in liquid nitrogen with a mortar and pestle. Five volumes of a buffer containing 50 mM Tris-HCl (pH 7.5), 0.05% (w/v) cysteine, 0.1% (w/v) BSA, were added just as the last trace of liquid N₂ disappeared. The thawed mixture was then ground and centrifuged at 20000 g for 15 min. The supernatant was used for spectrophotometric analysis.

Cytosolic APX (L-ascorbate: hydrogen peroxide oxido-reductase, EC 1.11.1.11) activity was measured according to Locato et al. (2009) by following the H₂O₂-dependent oxidation of ASC at 290 nm in a reaction mixture containing 0.1 M Tris-acetate buffer, pH 6.4, 350 μM ASC, 170 μM H₂O₂, 50–100 μg protein. The non-enzymatic H₂O₂-dependent oxidation of ASC, as well as the oxidation of ASC, not dependent upon H₂O₂ addition, was subtracted.

Dehydroascorbate oxido-reductase (glutathione: DHAR, EC 1.8.5.1), MDHAR (NADH: monodehydroascorbate radical oxido-reductase, EC 1.6.5.4), and CAT (hydrogen peroxide: hydrogen peroxide oxido-reductase, EC 1.11.1.11) were assayed according to Paradiso et al. (2012). Peroxidase activity (POD EC 1.11.1.7) was measured following the oxidation of 3,3′,5,5′-Tetramethylbenzidine (TMB) at 652 nm (𝜀 = 26,9 mM^-1 cm^-1), according to Ferrer et al. (1990).

Protein content was determined according to Bradford (1976) using bovine serum albumin as standard. All enzyme activities were measured using a Beckman (Fullerton, CA, USA) DU 7000 spectrophotometer.

Total RNA was isolated from Arabidopsis cells using the RNeasy plant minikit (QUIAGEN S.p.A., Milan, Italy) according to the supplier’s recommendations. Residual DNA was removed from the RNA samples using a DNA-free kit (AMBION, Inc., Austin, TX, USA). Synthesis of cDNA was performed from 2 μg total RNA with 10 μM random primers (AMERSHAM Biosciences Europe GMBH, Milan, Italy), utilizing an Omniscript Reverse Transcriptase kit (QUIAGEN S.p.A., Milan, Italy) according to the supplier’s recommendations. PCR reactions were performed with specific primers for APX1 (X59600, 5′-GGACGATGCCACAAGGATAG-3′ and 5′-GGTTGCGATTTGAACACAT-3′); APX2 (X98275, 5′-ATTGCCGTTAGGCTTCTTGA-3′ and 5′-TACCAACCGACAAGGCTCTT-3′); APX6 (AV555486, 5′-CTGCTGGTGTGCTTCGTTTA-3′ and 5′-TTGAAAAACCATGGACGTCA-3′) and 18S rRNA (AJ236016, 5′-CATGATAACTCGACGGATCG-3′ and 5′-GAAGGCCAACGTAATAGGAC-3′). 18S rRNA was used as an internal control in order to normalize each sample for variations in the amount of initial RNA. For semi-quantitative RT-PCR, the cycle number in the linear range was empirically determined. These were analyzed on 1.5% agarose gel containing 0.5 μg mL^-1 ethidium bromide.

Experiments were repeated five times and a statistical analysis was performed. Data were analyzed by one-way analysis of variance (ANOVA).

At different time intervals cell samples of Arabidopsis cell cultures subjected to anoxia and re-oxygenation were taken and analyzed for cell viability. Since no significant differences were observed between control and treated cells, cell viability was not affected by anoxia followed by re-oxygenation (data not shown).

The production of the main reactive species in Arabidopsis cell cultures was evaluated during normoxia, anoxia, and re-oxygenation by using fluorescent probes able to detect ROS, H₂O₂, and NO. Only the spotted red fluorescence, due to the autofluorescence of chloroplast chlorophylls was observed in cells incubated in the absence of the fluorescent probes as control (data not shown).

Reactive oxygen species production was detected in Arabidopsis cell cultures under different oxygen levels by using carboxy-H₂DCFDA (Figure 1). This probe, when deacetylated by intracellular esterases and oxidized by ROS, emits bright green fluorescence. In normoxic conditions (T0), very few cells showed faint fluorescence as defined single spots, likely due to the basal metabolism. During anoxia (T1 and T2), a diffuse green fluorescence was observed in an increasing number of cells indicating that ROS increase affected the whole cell, thus suggesting the occurrence of an oxidative stress, an usual event under different stress conditions. During re-oxygenation (T3 and T4), a reduction of both fluorescent cell number and fluorescence intensity was observed. Twenty hours after the end of anoxia (T4), cells seemed to restore completely the normoxic conditions being ROS detectable as spotted green-fluorescence.

Among ROS, the production of H₂O₂ was reported to increase in several abiotic stress conditions including oxygen deprivation (Blokhina and Fagerstedt, 2010). This ROS is of particular relevance for its relative stability under different physiological conditions and its capability to cross biological membranes. Indeed, it is well known that alterations in H₂O₂ are part of the redox signaling pathways activating defense responses in all aerobic organisms. H₂O₂ accumulation was then investigated by using 1,2,3-dihydrorhodamine (DHR), a probe which in the presence of H₂O₂ is oxidized to its fluorescent form R123 (Hensley et al., 2003).While DHR passively diffuses across plasma membrane, the charged fluorescent rhodamine-123 cannot cross being trapped in cellular compartments (Kinsey et al., 1987; Henderson and Chappell, 1993). Confocal images (Figure 2) show that, in normoxic conditions (T0), only few Arabidopsis cells displayed green fluorescence, due to the basal metabolism. Under anoxic conditions, H₂O₂ levels strongly increased. It is worth noting that during anoxia cell boundaries appeared to be particularly involved in H₂O₂ production (T1, T2, and T2m). Several enzymatic systems are responsible for O₂^-/H₂O₂ production at plasma membrane/cell wall interface. Among these, NADPH oxidase and apoplastic superoxide dismutase are well known enzymes participating in ROS overproduction in a plethora of environmental biotic and abiotic stresses, including low oxygen (Pucciariello and Perata, 2016). Specific PODs and amine oxidases also contribute to cell wall O₂^-/H₂O₂ production at least in some cases of stress conditions (Hurkman and Tanaka, 1996; Angelini et al., 2008). At the end of anoxic stress (T2), most Arabidopsis cells (about 70–80%) appeared green or yellow-green fluorescent. An increase of H₂O₂ during anoxia is coherent with our previous results indicating a rise for H₂O₂ released in the culture medium during anoxia (Nisi et al., 2010). During re-oxygenation (T3 and T4) H₂O₂ production decreased, only few cells being green fluorescent. The cellular compartment involved in H₂O₂ production also seems to be different being the plasma membrane/cell wall interface probably less relevant and intracellular production more relevant during re-oxygenation than in anoxia. This is consistent with the recovery of mitochondrial electron flow, previously blocked by oxygen deprivation and with a role for H₂O₂ in metabolic signaling within cells.

A tight interplay between ROS and reactive nitrogen species has been reported in stress signaling; NO has also been reported to be induced by anoxic conditions (Stöhr and Stremlau, 2006). NO production was then analyzed in Arabidopsis cells by CLSM using DAF-2 DA, a probe highly specific for NO (Kojima et al., 1998; Nakatsubo et al., 1998). Because of the short half-life (5–15 s) of NO, the detected fluorescence indicates the amount and the cell localization of NO production at a given time (Figure 3).

In normoxia conditions (T0), a basal level of NO production was detected as green fluorescent spots only in few scattered cells (Figures 3a,a₁); fluorescence was clearly visible in round shaped endocellular organelles; according to merge picture, almost no production was evident in chloroplasts (Figures 3f,f₁). The subcellular site of NO synthesis in plants is still under debate. Among the subcellular compartments suggested to be responsible for such biosynthesis, peroxisomes are probably the best-characterized (Corpas et al., 2013). However, under hypoxic conditions, NO production was also suggested to occur in other compartments such as the cytosol and organelles including mitochondria, which were shown to be a significant source of NO in various organisms including plants (Gupta et al., 2011). During the anoxic stress, a great increase of NO production was observed. It is worth noting that at 2 h anoxia, most cells appeared green fluorescent. At this early stage of anoxia (T1), Arabidopsis cells showed a diffuse green fluorescence due to NO within the whole cell volume (Figures 3b,b₁,g,g₁). The pre-incubation with the NO scavenger 2-(4-Carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide confirmed that green florescence was actually due to NO (data not shown). In T2, ∼60–80% of Arabidopsis cells displayed a positive reaction to the probe DAF-2 DA (Figures 3c,c₁,h,h₁). During the early recovery period 2 h after re-oxygenation NO was still produced in a relevant number of cells (T3, Figures 3d,d₁,i,i₁) even if at lower levels compared to T1. After 20 h of re-oxygenation (T4), cells restored the baseline conditions of NO production with very few cells showing faint green fluorescence (Figures 3e,e₁), which was evident as small dots (Figures 3j,j₁). Taken together these results indicate that oxygen shortage induced an oxidative burst. ROS and NO production seemed to be particularly high during the 1st hours of anoxia. During re-oxygenation, ROS, and NO were still produced. Under our experimental condition, almost 20 h were required in order to recovery redox homeostasis.

The transient increase of ROS, H₂O₂, and NO in Arabidopsis cultured cells during oxygen deprivation was likely able to trigger cell response for allowing the cells to overcome the anoxic stress. Consistently, in spite of such increases, no decrease in cell viability was induced.

In plant cells a complex network of redox metabolites and enzymes guarantees redox homeostasis. This network includes several ROS scavenging enzymes (CAT, PODs) and a network of antioxidant molecules, such as ascorbic acid, glutathione, α-tocopherol, that interact with ROS directly or in reactions carried out by antioxidant enzymes such as APX (Noctor and Foyer, 1998; Apel and Hirt, 2004). In addition, a whole array of enzymes is needed for the regeneration of the active forms of the antioxidant molecules (MDHAR, DHAR, and GR).

The importance of the ASC-glutathione system for the detoxification of hydrogen peroxide has been well characterized in different stress conditions (Locato et al., 2009; de Pinto et al., 2015) but little detailed information is available about its action in anoxic stress.

The behavior of ASC pool during anoxia indicates that no significant alteration occurred when the total ASC plus DHA is considered (Figure 4A). In contrast, the redox state (ASC/ASC + DHA) was severely affected by the variations of oxygen concentration. After 2 h of anoxia (T1), a significant increase of the oxidized form (DHA) was observed, resulting in a decrease of the redox state (0.82 ± 0.15 and 0.11 ± 0.042 at T0 and T1, respectively) that remained unchanged up to the end of anoxia (T2). This confirms the strong redox impairment occurring during anoxia. After 2 h of re-oxygenation (T3), however, the total content of ASC increased significantly, reaching almost doubled values compared to T0 cells. T4 cells showed ASC levels and redox state fully comparable to those of T0 cells (de Pinto et al., 2006; Munné-Bosch et al., 2013).

The anoxic treatment caused a significant decrease of total glutathione content during the first 2 h of anoxia (T1) (Figure 4B), while glutathione pool increased again to the T0 value in the following 2 h of anoxia. Differently from ASC, GSH redox state remained unchanged during oxygen concentration changes, with a ratio of more than 90% in reduced form. After re-oxygenation (T3), GSH content increased over threefold: the increase of ASC and glutathione pool, occurring at this stage, emphasized that the cells were responding to an extraordinary stress situation by enhancing their redox molecules. At T4 GSH returned to control values, similarly to what occurred for ASC, thus showing that cells were able to restore the physiological redox balance.

The levels of the lipophilic antioxidant metabolite α-tocopherol were also measured during normoxia, anoxia, and re-oxygenation (Figure 4C). In these experimental conditions, α-tocopherol showed a slight but significant increase at T2 phase and a slow decrease after re-oxygenation until reaching values similar to control cells (T4). In plants, α-tocopherol is known to increase in several environmental constraints involving oxidative stress (Munné-Bosch, 2005) thus an increase during anoxia could be expected. On the other hand, since ASC is involved in the process of regeneration of tocopherols, such regeneration during anoxia might likely contribute to the observed decrease in the ASC/DHA redox state.

To maintain the redox state and to regenerate antioxidant molecules in their active form, several enzymes act to support the antioxidative defense. The recycling enzymes of ASC-GSH cycle were tested in order to follow their behavior in Arabidopsis cell cultures at different oxygen concentrations (Figure 5). MDHAR and DHAR, the enzymes responsible for the reduction to ASC of MDHA and DHA, respectively, did not change significantly in response to both limited oxygen availability and following re-oxygenation. The activity of GR, enzyme responsible for glutathione reduction, significantly decreased during anoxia and gradually reached control values after re-oxygenation. It is worth noting that in spite of GR decreases during anoxia, the redox state of GSH pool remained high, thus suggesting that GR activity was overabundant.

To obtain information of the global H₂O₂ removal capability of Arabidopsis cells in response to oxygen concentration changes, the activity of APX, CAT, and POD was also measured (Figure 6). POD and CAT activity drastically decreased during anoxia to about 50% of the control value already after 2 h of oxygen deprivation; re-oxygenation induced a quick restoration of CAT activity while POD activity was completely restored only at the end of the experiment.

Among the APX isoenzymes, the activity of cytosolic ones (cAPX) was determined: under our experimental conditions, photosynthesis was not active (data not shown). Moreover, cytosol has been reported to act as key site of redox signaling integration (Noctor and Foyer, 2016). Anoxia resulted in a significant 40% decrease of cAPX activity only at T2. The high activity of this H₂O₂ scavenging enzyme during the first period of oxygen deprivation supports a key role for this enzyme in overcoming oxidative damage. During re-oxygenation, cAPX activity rapidly increased. This result is coherent with other evidences obtained when oxygen-deprived roots of wheat seedlings were re-aerated (Biemelt et al., 1998). Three genes encoding for cAPXs are present in Arabidopsis (Noctor et al., 2016). To test whether alteration in cAPX occurring during anoxia and re-oxygenation was due to gene expression changes, the transcript accumulation of these three genes was analyzed (Figure 7). The sequences of RT-PCR amplifications were compared to predicted mRNA sequences, confirming the correspondence with Arabidopsis genes (data not shown).

During anoxic treatment, the three cAPX genes exhibited a similar pattern of transcript accumulation, showing no significant changes in their expression. Re-oxygenation induced an immediate increase in the transcript accumulation that was particularly high for APX2. This is consistent with the pivotal role of APX2 in responses to oxidative stress, since this isoenzyme is considered a stress marker and its expression is induced by several kinds of stresses (Gallas and Waters, 2015). Redox-regulated transcription factors have been characterized. The promoter of Arabidopsis APX1 and APX2 contains a heat shock element binding a heat shock factor sensitive to endogenous H₂O₂ production (de Pinto et al., 2015). In rice, an increase of H₂O₂ levels enhanced the expression of two cytosolic APX genes (Li et al., 2015). For all the three analyzed cAPXs the expression values returned comparable to the control after 20 h of re-oxygenation. The increase in cAPX gene expression observed when the anoxic cells were again exposed to oxygen might be related to the increase of enzyme activity occurring during re-oxygenation. On the other hand, no correlation was observed between the decrease of cAPX activity occurring at T2 and gene expression changes. It has been suggested that post-translational modifications are rapid regulators of enzyme activity. Most of these post-translational modifications are redox regulated and involve NO or ROS, such as carbonylation of specific amino acids, thiol – disulphide transition and S-nitrosylation (de Pinto et al., 2013; Waszczak et al., 2015). Due to the strong oxidative environment occurring during oxygen deprivation, it is likely that redox post-translational modifications are responsible for changes in enzyme activities, even if more in deep studies are required for verifying this hypothesis.