The physiological response of the marine platyhelminth Macrostomum lignano to different environmental oxygen concentrations

Marine meiofauna colonize the upper sediment layer of the ocean, which, especially in coastal and intertidal sediments, can be highly variable with respect to oxygenation. On the one hand, wave action and coastal currents often lead to mechanical mixing of oxygen-rich surface water into the upper loosely packed sediment layers. Relatively compacted sediment surfaces in shallow waters are often densely colonized by benthic microalgae that cause daily peaks of photosynthetic oxygenation in the upper 1–2 cm of sediment surface. On the other hand, coastal sediments contain high amounts of organic matter from decaying macroalgae or terrestrial and riverine coastal runoff. The chemical and microbial oxygen demand in these sediments is high so that oxygen that diffuses downwards is rapidly consumed. This produces steep sedimentary redox gradients between the upper, sometimes even hyperoxic and the lower suboxic to anoxic layers, which often span no more than 10–50 cm, below which the sediment becomes chronically anoxic and sulfidic (rich in hydrogen sulfide, H₂S) (Fenchel and Finlay, 2008). These sediments are colonized by highly diverse macrofauna and meiofauna communities, the composition of which varies depending on sediment stability, grain size composition, organic matter content and oxygenation. Whereas macrofauna organisms often build sedimentary burrows that they oxygenate through irrigation with oxygen-rich surface waters, meiofauna organisms move between the sand grains, seeking optimal positioning between too high and too low environmental oxygenation according to their metabolic and nutritional requirements. In very few cases were marine meiofauna found to actively establish suitable environmental oxygen levels in their closest environment between the sediment grains (Corbari et al., 2005). Much more often it was shown that the meiofauna stratify in response to the sedimentary oxygen gradient (Corbari et al., 2004) and different groups of meiofauna have been described as oxophilic, microoxophilic and thiobiotic according to their preferred positioning in the small-scale chemical and redox gradients within the sediment column and around macrofauna burrows (Fenchel and Finlay, 1995; Morrill et al., 1988). This behavior precludes not only that meiofauna species have different metabolic strategies to cope with low and variable environmental oxygenation but also that they would have the means to sense oxygen (and hydrogen sulfide) and react flexibly to changing environmental oxygenation. These reactions include vertical migrations within the sediment column, often away from the too high surface oxygenation (Corbari et al., 2005), and/or adjustment of metabolic rate to cope with reduced oxygen availability. However, soft-bodied meiofauna worms, such as nematodes and platyhelminths, generally deal with fluctuating tissue oxygenation and changes in environmental redox state because of their small size and fast equilibration time. Whereas several studies have assessed the behavioral (Rosenberg et al., 1991; Tinson and Laybourn-Parry, 1985) and ecological responses to fluctuating oxygenation (Wetzel et al., 2001), studies of the physiological response to fluctuating chemical and oxygenation gradients are scarce.

A seminal paper by Morrill and colleagues (Morrill et al., 1988) reported antioxidant activities to be better established in thiobiotic than in oxybiotic meiofauna, giving rise to the idea that (i) normoxic oxygen levels could already be ‘hyperoxic’ for these low oxygen-adapted species, and (ii) the oxidation of toxic H₂S could give rise to the formation of reactive oxygen species (ROS) in thiobiotic, low oxygen-adapted meiofauna.

In the present study, we investigated the physiological effects of hyperoxic and anoxic exposure conditions in the interstitial marine flatworm Macrostomum lignano (Rhabditophora, Macrostomorpha) (Ladurner et al., 2005). Macrostomum lignano has been established as a new model in sexual selection (Schärer et al., 2005) and evolutionary and developmental studies (Egger et al., 2006; Morris et al., 2006), but especially in stem-cell research (Bode et al., 2006; Pfister et al., 2007) and the role of these stem-cells in the process of aging (Mouton et al., 2009).

Platyhelminths are transparent and, thus, elegant model organisms to study cellular and subcellular processes in vivo. Using ‘live imaging’ techniques we provide a first set of data regarding the physiological responses of individual flatworms to extreme states of environmental oxygenation and, further, on the effect of anoxia and subsequent reoxygenation on ROS-forming processes, mitochondrial membrane potential (ψ_m) and tissue pH. We developed a technique to record oxygen partial pressure (P_O2)-dependent oxygen consumption rates using batches of 20 flatworms in order to describe their metabolic response to variable environmental oxygenation.

In this study we used non-invasive optical techniques to understand how an anoxia-tolerant animal responds to high, low and fluctuating oxygenation, especially with respect to ROS formation, mitochondrial functioning and maintenance pH homeostasis during anoxia–reoxygenation, a detrimental situation in many human pathologies.

Cultures of M. lignano (DV-1 line) were reared at the Alfred Wegener Institute (Bremerhaven, Germany) in Petri dishes with Guillard's F/2 medium at room temperature (RT, 20°C) and fed weekly with the diatom Nitzschia sp. For experimentation, however, a sterile medium composed of seawater filtered through a 0.2 μm Whatman filter and supplemented with 15 mmol l⁻¹ Na-Hepes (NH-FSW) was used. Buffering was necessary to avoid pH changes in the medium when setting anoxic or hyperoxic conditions.

Two different treatments were considered for the respiration measurements: normoxia (21 kPa) and anoxia (0 kPa) followed by reoxygenation. For the latter, animals were kept for 1.5 h under anoxic conditions (in a gas-tight glovebox that was equilibrated with 100% N₂ and immersed in an anoxic medium, prepared by flushing with 100% N₂) and later reoxygenated by bubbling with air for an additional 1.5 h. Prior to the respiration measurements, experimental animals were isolated from the cultures and maintained without food for ca. 20 h in NH-FSW. After this acclimation time, three groups of 20 animals were transferred with a micropipette to three individual wells of a glass microtiter plate (Mikroglas Chemtech, Mainz, Germany) previously equipped with oxygen sensor spots (SP-PSt3-NAU-D5-YOP, Precision Sensing, Regensburg, Germany) glued onto the bottom of each well. An additional well was maintained as a control, and contained only NH-FSW and the sensor spot. Each of the wells was filled completely with NH-FSW to its maximum capacity, which was 100±0.5 μl, to avoid the formation of air bubbles when sealing the wells. Each of the four wells was sealed with a coverslip and the complete surface of the microtiter plate was additionally covered with a layer of auto-adhesive Armaflex (Armacell Enterprise, Münster, Germany). Additional pressure was maintained on the complete surface of the plate in order to ensure air-tight sealing. Measurements were carried out with a four-channel fiber-optic oxygen meter (Oxy-4) and non-invasive oxygen sensors (Precision Sensing), which were daily calibrated following the manufacturer's description. Data were recorded at 15 s intervals and the experiments were stopped when the oxygen was completely consumed from each of the animal chambers, which usually took around 16 h. All experiments were conducted at RT. Data corresponding to the first 30 min following the start of the experiment were discarded in order to avoid interference from stress related to the manipulation of the worms. Respiration rates were expressed as nmol O₂ l⁻¹ h⁻¹ individual⁻¹. The critical oxygen partial pressure (P_c) (Tang, 1933) was calculated using the equation of Duggleby (Duggleby, 1984).

In this study, intracellular pH (pH_i), ψ_m, mitochondrial density and ROS concentration were measured by ‘live imaging’ techniques, involving the application of specific dyes and in vivo visualization using a Leica TCS SP5II confocal microscope (Leica Microsystems CMS, Wetzlar, Germany) and a CCD camera system (Visitron Systems, Puchheim, Germany).

Four treatments were considered in our study: control (21 kPa), anoxia (0 kPa), hyperoxia (40 kPa) and anoxia followed by reoxygenation. Experiments were carried out in a gas-tight glovebox, which was equilibrated with either 100% N₂ (anoxia) or 40% O₂/60% N₂ (hyperoxia) using a Wösthoff gas-mixing pump (Wösthoff, Bochum, Germany). Anoxic medium was prepared by flushing with 100% N₂, while hyperoxic NH-FSW medium was obtained by flushing with a 40% O₂–60% N₂ mixture; flushing for 30 min was enough for equilibration in all cases, as confirmed by measurement of the oxygen content using optical oxygen sensors and the Presens Oximeter (see above).

Treatments were conducted with batches of 10 animals that were manipulated at the same time in 2 ml Utermöhl chambers for 1.5 h. The microscopic chamber in which the animals were anoxically incubated was sealed while still inside the glovebox and under the gas stream. The chamber was then transferred to the confocal microscope. For the anoxic treatment, it should be noted that we cannot exclude some minor reoxygenation; thus, we decided to use the term ‘near-anoxia’ for the treatment that was aimed to be strictly anoxia. At the end of the incubation, a fluorophore was added to the animals in normoxic, anoxic (severely hypoxic) and hyperoxic treatments.

For the anoxia–reoxygenation treatment, the worms were incubated in absolute anoxia; the chamber was subsequently opened and medium and animals were allowed to reoxygenate for an additional 1.5 h. After this time, the dye was added to the reoxygenation treatment. All dyes were used individually for separate batches of animals in order to avoid any possible interference among dyes. The list of the different dyes used in the present study along with their mechanisms of function is given in Table 1, while incubation and analysis conditions for each of these dyes are detailed in Table 2.

For both of the pH-sensitive dyes used in the present study (Ageladine-A and BCECF), further calibration was needed in order to assign numerical pH values to the experimentally recorded intensities/ratios. For Ageladine-A (Fujita et al., 2003), this was done following a previously described method (Bickmeyer et al., 2008; Bickmeyer, 2012) in which the average values obtained for each of the treatments are expressed as a proportion of the control values, and by assuming that non-specialized cells under these control conditions would present a putative pH_i of 7.3–7.4. In contrast, as BCECF fluorescence intensity at 490 nm linearly increases between pH 6.4 and 7.8 (Silver, 2003), BCECF results were converted to pH_i by constructing a calibration curve using the nigericin technique (nigericin sodium salt, 5 mg, N7143, Sigma-Aldrich, St Louis, MO, USA) described previously (Thomas et al., 1979).

Fluorimetric analysis of the response of the animals to the treatment conditions was carried out in sealed microscope slides using a Leica TCS SP5II confocal microscope. For validation of the pH_i results, further analyses were carried out using a wide-field fluorescence microscope Zeiss Axiovert-10 (Zeiss, Jena, Germany; equipped a CCD camera system, Visitron) and applying BCECF staining. For analysis, animals were anesthetized in a 2:1 mixture of 7.14% MgCl₂· 6H₂O and NH-FSW (Pfannkuche and Thiel, 1988). One picture per individual and per photomultiplier tube (PMT), when ratio calculation was required, was taken for each of the animals using a 10× objective lens. In order to avoid photobleaching, a short period (<5 s) of low resolution (256×256 pixels) live scanning was applied for focus adjustments and afterwards only one single scan (512×512 pixels) was run of each individual. Autofluorescence was suppressed by adjusting the threshold, and phototoxicity was minimized by the multiphoton laser for low wavelength scans.

Image analysis was carried out using Leica LAS Lite software (Leica Microsystems). Quantification was carried out by plotting five transects per animal (always perpendicular to the longitudinal axis of the animal and in all cases selecting the areas of highest intensity; Fig. 1). The maximum value for each transect was obtained and the values of the five transects were averaged for each individual.

Experiments were repeated a minimum of two times (if results were identical) and up to four times, in order to confirm the differences between treatments. Data from the different replicas were pooled. However, as dye uptake by the animals varied on different days, the confocal settings needed to be adjusted in order to optimize the quality of the images taken. Thus, data normalization was required prior to pooling. Data were normalized by expressing all values within one experimental replica (10 worms) as a percentage of the maximum recorded value in that experiment. The maximum value was set to equal 100%. Statistical analyses were performed using SPSS 15.0 (SPSS Inc., Chicago, IL, USA). The resulting data were compared using ANOVA (followed by Student–Newman–Keuls a posteriori multiple comparison test) when data complied with the assumptions for parametric analyses. When this was not the case, Kruskal–Wallis tests were conducted.

For the normoxic treatment, P_O2-dependent respiration rates were analyzed in a total of 10 pools, each consisting of 20 animals. Fig. 2 shows that respiration rates decreased rapidly (by 70%) between 19 and 12.5 kPa in the chamber. Below this P_O2, M. lignano starts to regulate respiration down to a low critical P_O2. Thus, for M. lignano, a first P_c was established at 13 kPa (P_c1), which marks the point at which the organisms are switching from conformity to regulation. Between 12.5 and ~3 kPa, the respiration rates of the worms were constant at around 0.064±0.001 nmol O₂ l⁻¹ h⁻¹ individual⁻¹. A second P_c appears at 1.36±0.32 kPa (P_c2), where respiration rates started to decrease again with declining P_O2. Thus, the analysis reveals a high and a low P_c above which (P_c1) and below which (P_c2) the worms switch from regulating to conforming P_O2-dependent respiration. The same pattern was observed for the animals that were previously maintained under anoxic conditions and subsequently reoxygenated. However, in the interval at which animals maintained constant respiration rates, values were significantly higher for the anoxia–reoxygenation group (0.083±0.001 nmol O₂ l⁻¹ h⁻¹ individual⁻¹) (F=458.23; P<0.001).

Normoxic and hyperoxic animals moved rapidly in the Petri dishes with a behavioral characteristic best described as unidirectional burst swimming. Consequently, for microscopic analysis at these high oxygenation levels, the worms had to be anesthetized with MgCl₂. In contrast, movement was drastically reduced in anoxia and often the worms started to swim in circles. Many worms were completely immobilized and displayed only epithelial cilia movements. Movements were completely reactivated by reoxygenation, with animals often reaching normoxic levels of activity after 1.5 h under reoxygenated conditions.

Staining with the pH-sensitive dye Ageladine-A (Fig. 3A; supplementary material Fig. S1A) indicated that the tissue pH levels in anoxic worms or those exposed to anoxia–reoxygenation were significantly more alkaline than those of normoxic and hyperoxic animals (ANOVA: F=10.38; P<0.001). Further estimation of the pH values (using the values of normoxic animals as a reference and assuming that non-specialized cells have a pH of 7.4) indicated a pH increase by ~0.10 units under near-anoxia and anoxia–reoxygenation conditions.

Staining with BCECF (supplementary material Fig. S1B) corroborated the general pattern obtained with Ageladine-A in which anoxic and reoxygenated individuals were significantly more basic than the normoxic and hyperoxic treatment groups (F=3.70; P<0.05) (Fig. 3B). As these results contradicted all our assumptions that animals exposed to near-anoxia should become, if anything, more acidic, we applied yet another approach with the wide-field fluorescence imaging microscope and BCECF. Again, animals under near-anoxia and anoxia–reoxygenation had the most alkaline values (supplementary material Fig. S2) (showing an average increase of 0.07 units), whereas the normoxia- and hyperoxia-treated worms had a tissue pH of 7.46.

MitoTracker staining indicated that most mitochondria were located in the animal's body wall in epithelial and muscle cells, as well as around the mouth area (Fig. 4). MitoTracker Green FM (supplementary material Fig. S3A) stained different treatment groups with different intensities (K=54.09; P<0.000). Animals kept under hyperoxic conditions (40 kPa) were stained with the highest intensity, theoretically suggesting maximum mitochondrial density in hyperoxic worms, followed by individuals under normoxic conditions, which displayed less intensive staining (Fig. 5A). Animals kept in anoxia had the lowest staining values with MitoTracker green, and individuals that were maintained in anoxia and subsequently reoxygenated returned to the staining intensity recorded in normoxia.

To determine whether the significant differences between differently oxygenated groups were associated with oxygenation-dependent capacities for dye uptake, a second experiment was performed in which the animals were stained prior to exposure to the different oxygenation treatments. With this approach we obtained exactly the same group-specific pattern as in the first experiment in which the dye was applied following incubation at different oxygen levels. Therefore, we used MitoTracker Deep Red 633 in a third approach (supplementary material Fig. S3B), which did not produce a clear group-specific pattern. Pooled data from the four MitoTracker Deep Red 633 replicas yielded no significant differences among treatments (K=6.98; P=0.072) (Fig. 5B). High resolution confocal imagery confirmed staining of individual mitochondria in M. lignano cells (Fig. 6).

Staining with the dye JC-1 (supplementary material Fig. S3C) indicated significant differences in the chemiosmotic H⁺ potential (Δψ_m) across the mitochondrial inner membrane among treatments (F=15.14; P<0.001) (Fig. 5C). The smallest potentials were measured in animals exposed to normoxia and in those that were reoxygenated after 1.5 h of anoxic incubation. Animals exposed to hyperoxic and anoxic media showed significantly more red/green staining, indicating an elevated membrane potential, i.e. a stronger H⁺ gradient.

Superoxide anion (O₂•⁻) concentrations were assessed with dihydroethidium (DHE) (supplementary material Fig. S4A). DHE staining was similar in animals exposed to anoxic, normoxic and hyperoxic conditions. Animals that were reoxygenated after a period of 1.5 h in anoxia had significantly increased DHE staining compared with all other groups (K=15.77; P<0.001) (Fig. 7A). Formation of other ROS species aside from O₂•⁻ (H₂O₂, HOO• and ONOO⁻) in M. lignano cells was assessed using C-H₂DFFDA (supplementary material Fig. S4B), which again showed important differences between treatments (K=80.82; P<0.000). The staining intensity decreased in the order anoxia–reoxygenation=hyperoxia>normoxia>near-anoxia (Fig. 7B), indicating ROS formation to decrease between hyperoxia and anoxia as expected.

To scrutinize for the causes of the high DHE signal in anoxically incubated worms that should theoretically not be able to form O₂•⁻, we applied a superoxide dismutase (SOD) solution (Sigma S-2515) and the SOD inhibitor sodium diethyl-dithiocarbamate trihydrate (SDT; Sigma D-3506) to anoxic animals stained with DHE (Fig. 8A). The first results with neither SOD nor SDT addition again led to significant DHE staining (above the background signal) in anoxically incubated specimens, although the values were significantly lower than those obtained with normoxic animals. SOD addition to anoxically incubated animals significantly reduced DHE staining, as O₂•⁻ radicals were converted to H₂O₂ at higher rates. In contrast, addition of the SOD inhibitor SDT caused a 2-fold higher DHE fluorescence in anoxic animals compared with normoxically exposed animals, because fewer O₂•⁻ radicals were converted to H₂O₂. Around 90% of the animals died under anoxic exposure with addition of SDT. After performing these tests to indirectly confirm the specificity of DHE for O₂•⁻ detection, we assessed the effect of SOD addition on DHE and C-H₂DFFDA staining in an experiment with both dyes (Fig. 8B). Addition of SOD to the medium of anoxically incubated worms once again significantly reduced O₂•⁻ in the DHE-stained worms compared with the normoxic control group. In contrast, SOD addition significantly increased C-H₂DFFDA staining, suggesting higher H₂O₂ concentrations in anoxically incubated SOD-supplemented animals than in normoxic controls.

In the present study we used living M. lignano individuals for the application of live imaging techniques. The transparency of this organism allowed an adequate and satisfactory measurement of the physiological parameters to be taken into account. Moreover, its small size also favored a fast and complete dye uptake, making live imaging with the whole animal possible.

Respiration measurements in M. lignano cannot be conducted with single animals, and instead require batches of 20 specimens maintained in small volumes (90–100 μl) of liquid. As the respiration rates are extremely low, measurements need to be conducted in small glass or Plexiglas chambers without linings made of Teflon, which is known to be an oxygen binder.

Exposing M. lignano to decreasing oxygen concentrations during the respiration measurements showed that the worms maintain constant consumption rates between 3 and 12.5 kPa. This could indicate that this is the species' optimum range of respiration and the animals are actually regulating oxygen consumption, most likely by adjusting ciliary beat frequency. The velocity of ciliary pumping movements may determine the water flow over the body surface, which, in this species, also represents the respiratory epithelium. We could not quantify ciliary beat frequency in the present study; however, we did not observe remarkable differences in cilia movements under different oxygen concentrations. Although we cannot reject the hypothesis that animals are actively regulating ventilation through cilia beating, we should consider the possibility that such regulation does not exist and that, in fact, the results are driven by mitochondria saturation above 3 kPa and that the activity of an alternative oxidase could be responsible for the increase in O₂ consumption above 12.5 kPa. Regardless of the the reason, the range of constant respiration characterizes M. lignano as aerobic hypoxia-tolerant species. Similar P_O2-dependent respiration patterns have been reported for other marine meiofauna such as oligochaetes (Giere et al., 1999) and the authors suggested this to be an important prerequisite for successfully inhabiting intertidal sediments with microscale spatial variations in the oxygen concentration.

Animals that were reoxygenated after enduring a 1.5 h period under anoxia showed a 30% increase in their respiration rates, probably due to the stress of the procedure.

Δψ_m, and the way the ψ_m is affected by different oxygen concentrations, has mainly been addressed in mammalian cell cultures. For M. lignano, the JC1 measurements under normoxia and hyperoxia suggest that the increase in oxygen consumption is due not to mitochondrial uncoupling (opening of the mitochondrial membrane transmission pore) but to a more intensive respiration, electron flow, which causes increased Δψ_m in hyperoxically exposed worms. In contrast, mitochondrial uncoupling and a reduction of Δψ_m under hyperoxic conditions has been observed in mammalian microvascular cells (Sastre et al., 2000) and pneumocytes (Guthmann et al., 2005), which impressively highlights the difference between oxyconforming flatworms and oxyregulating mammalian cells. We did not observe aggregation of worms under normoxic/hyperoxic conditions indicating collective re-breathing as observed in ostracods when facing too high oxygen levels (Corbari et al., 2005). Indeed, a more thorough investigation of the migration or aggregation behavior of M. lignano within oxygenation structured micro-environments is warranted (Fenchel and Finlay, 2008). However, we did observe that the locomotory activity of the worms was conspicuously higher in normoxic and hyperoxic treatments, with no observable difference between the two oxygenation states. Thus, as in many other infaunal species (Abele et al., 1998), normoxic oxygen levels are already high for these plathyhelminths and their behavior indicates an attempt to reduce oxygenation in their sedimentary environment. This increased activity is fueled by the enhanced respiration above 12.5 kPa.

The C-H₂DFFDA measurements further indicate an increase in ROS (H₂O₂, HOO• and ONOO⁻) production compared with normoxic worms, whereas no effect of hyperoxia on O₂•⁻ concentrations was observed in the DHE measurements. A cross-experiment in which we added either SOD or the SOD inhibitor STD to the medium confirmed that the DHE signal depends primarily on the amount of O₂•⁻ in the cells. The measurements with ROS-sensitive fluorophores therefore indicate that under hyperoxia the worms' SOD activity converts excess O₂•⁻ to H₂O₂. This conversion is abolished by the SOD inhibitor STD, which killed the animals when they were exposed under hyperoxic conditions. Mitochondrial production of ROS increases linearly with increasing oxygen concentration (Turrens et al., 1982), and hyperoxia has been shown to elicit oxidative stress and antioxidant responses in marine model organisms (Abele et al., 1998; Lushchak and Bagnyukova, 2006). Interestingly, 12.5 kPa is also the level at which mitochondria of the mammalian lung start to release H₂O₂ at a faster rate (compared with lower oxygen levels), indicating that antioxidant defenses are starting to be overwhelmed. It seems a bit far-fetched to suggest a common upper P_c for such distinct systems as meiofauna species and human lung epithelia, but the comparison illustrates the wide applicability of the hypothesis that in very diverse systems, P_O2 is kept at very low levels to prevent hyperoxic ROS production.

The pH of the tissue in hyperoxia was shown to be slightly more acidic than that in animals kept under normoxic conditions. This can result from various mechanisms such as intensified hydrolysis of ATP and oxidative inhibition of the Na⁺/H⁺ antiporter under enhanced oxidative stress, as reported for brain cells (Mulkey et al., 2004) and for crustaceans (Abele-Oeschger et al., 1997). Furthermore, in fish and some crustaceans, hyperoxic shocks can result in a reduction of ventilation rates, which causes extracellular acidosis (Gilmour and Perry, 1994; Gilmour, 2001; Wheatly and Toop, 1989).

Under anoxic conditions, M. lignano showed drastically reduced movements, presumably an energy-saving behavior in response to the lack of oxygen. A similar response to anoxia has been reported for a wide variety of invertebrate species, like bivalves (e.g. de Zwaan and Wijsman, 1976) or crustaceans (e.g. Hervant et al., 1995).

Even though we did not obtain numerical Δψ_m results, our data indicate that M. lignano mitochondria mount a stronger H⁺ gradient under low oxygen concentrations than under normoxic conditions (Fig. 5). In normoxia, hypoxia-tolerant invertebrates have lower Δψ_m values than cold-blooded vertebrate and mammalian mitochondria (Abele et al., 2007; Brookes et al., 1998; Keller et al., 2004). Anoxic M. lignano also maintained high Δψ_m values as hyperoxic individuals, reinforcing the view that the species not only tolerates low oxygen concentration but also keeps up the mitochondrial proton gradient down to very low P_O2. Under severely hypoxic to anoxic conditions, <1.5 kPa, oxygen consumption rates started to decline as oxygen became limiting. Maintenance of the high proton gradient would then be due to reduced energy expenditure including the complete absence of swimming mobility so that energetic requirements can be estimated to be minimal. This leads to the conservation of a high H⁺ gradient because of a reduced phosphorylation activity and proton backflow at the ATPase.

However, anoxic animals also featured the most alkaline (least acidic) intracellular pH. This is paradoxical in the sense that for many organisms, cells and tissues, severe hypoxia has been demonstrated to lead to a decrease in cellular ATP, an increase of the free intracellular calcium concentration (e.g. Kristián and Siesjö, 1996) and acidification of extracellular and intracellular pH (e.g. Bickler, 1992). Only a few studies have observed hypoxic alkalization in cancer cells (for review, see Webb et al., 2011) and certain mammalian cells, in keeping with our results (Mitsufuji et al., 1995). Such an alkalization of the intracellular pH during anoxia could well be an adaptive strategy to avoid apoptosis, known to be associated with intracellular acidification (Lagadic-Gossmann et al., 2004; Matsuyama et al., 2000). But the question remains how such an increase of pH_i can be achieved. In cancer cells this has been associated with changes in the expression and/or activity of membrane transporters and ion pumps (see Webb et al., 2011). Other authors (e.g. Mitsufuji et al., 1995) suggested that the increasing pH values might be attributed to the activation of the Na⁺/H⁺ antiporter during hypoxia. As the Na⁺ gradient across the cellular membrane is maintained by the Na⁺/K⁺-dependent ATPase, the driving force for the proton export by the Na⁺/H⁺ antiporter is an energy (ATP)-dependent process. The Na⁺/H⁺-antiporter exports intracellular protons, thus increasing pH_i and preventing acidosis until the cellular stores of ATP are exhausted. We have, however, demonstrated that ψ_m and potentially also the electron flow are maintained down to very low oxygen concentrations. As ATP requirements are reduced through the reduction of locomotion, we assume ATP to be available for the upkeep of the Na⁺ gradient and therefore for stabilizing secondary active proton export during severe hypoxia, which would explain the less acidic values in our near-anoxia worms. As ATP concentration was not measured in the present study, this hypothesis cannot be corroborated. In future studies, ATP concentration and the involvement of membrane transporters in stabilizing intracellular pH should be tested by luminescence assays (ATP+APD) and using specific inhibitors for membrane transporters. Another factor that could theoretically contribute to the more alkaline pH values during periods of low oxygen concentration would be the decrease in ATP hydrolysis, which also involves H⁺ release, because of the reduction of the worms' muscular activity. Although we consider hypoxic alkalinization to be an important new finding based on live imaging techniques and confirmed by the use of two different pH-sensitive dyes, further investigations of this phenomenon are recommended. While no exact calibration in vivo has yet been developed for Ageladine-A, BCECF has a reported accuracy of 0.07 pH units (Franck et al., 1996), which is within the range shift that we detected in the present study. Further experiments are planned to see whether pH_i alkalinization during hypoxia in the worms is accompanied by an extracellular acidification as observed in cancer cells (Webb et al., 2011).

The tenet that mitochondrial ROS formation increases linearly with oxygen concentration in cells (Turrens et al., 1982) is also disputable for the worms. Whereas H₂O₂ formation was in agreement with this theory, O₂•⁻ formation follows a different pattern: M. lignano individuals under near-anoxia showed similar O₂•⁻ concentrations to normoxic and hyperoxic animals. Under conditions of strict anoxia, and thus in the absence of oxygen, ROS can definitely not be produced. In consequence, under so-called anoxia, minimal traces of oxygen must still have been present, enough to cause some O₂•⁻ formation in the worms. It is still under debate whether ROS can be produced under hypoxia (Hermes-Lima et al., 1998; Hermes-Lima and Zenteno-Savín, 2002) and some evidence of this happening came first from direct ROS measurements (Vanden Hoek et al., 1997), followed by detection through resulting DNA damage (Englander et al., 1999). These results can, however, also be explained by the low K_m for O₂ of mitochondrial nitric oxide synthase, which would lead to production of NO, estimated to be around 5–10% of the normal steady rate of NO production (Alvarez et al., 2003). This NO may bind to and inhibit cytochrome oxidase, causing an increase in its K_m for oxygen and, consequently, an increased reduction of the upstream electron transporters such as complex I and III, and therefore enhanced formation of O₂•⁻ under hypoxic conditions (reviewed by Turrens, 2003). Further, NO can react with superoxide to form toxic peroxynitrite, which can lead to all sorts of macromolecular damage, including DNA damage, and presumably also induce antioxidant systems. Based on the results obtained through the use of SOD and STD and their effect on anoxic individuals, we suggest that SOD might be inhibited during anoxia, which would additionally contribute to the high O₂•⁻ and the low H₂O₂ concentrations recorded under these conditions. For other invertebrate species such as clams, a decrease in SOD expression and activity has also been observed (Monari et al., 2005).

Mitochondria in M. lignano are mainly concentrated in muscular and epithelial cells, providing energy to the numerous cilia and for body contraction. Although the two MitoTracker dyes stained the same location, MitoTracker Green FM and MitoTracker Deep Red 633 gave significantly different results (Fig. 5; supplementary material Fig. S3) with respect to the comparison across treatments. Only with MitoTracker Green did we observe a decrease in staining under anoxia, which suggests fluorescence of this dye is negatively affected by anoxic exposure, even though it has been reported to be independent of membrane potential (Métivier et al., 1998). Some studies have demonstrated that this may not be true (Keij et al., 2000), and similar oxygenation-dependent patterns of JC-1 and MitoTracker Green FM in M. lignano strongly suggest that MitoTracker Green is indeed sensitive to Δψ_m. In this case, MitoTracker Green FM cannot be considered as a useful probe for mitochondrial density in experiments involving variable tissue oxygenation. This is further supported by the fact that MitoTracker Deep Red 633 staining was independent of the experimental oxygenation state and can thus be considered a better tool for determining mitochondrial mass (Haugland, 2002) in this kind of experiment.

In an attempt to establish the platyhelminth M. lignano as an applicable whole-animal model for physiological investigations, we determined its peculiar physiological response to anoxic and hyperoxic exposure at different levels of cellular functioning. The worms are hypoxia tolerant and maintain the mitochondrial proton gradient and presumably also ATP levels during at least 2 h of anoxic exposure. Oxyconforming respiration below 3 kPa apparently satisfies the worm's maintenance metabolism and prevents the onset of cellular acidosis, while mobility ceases altogether and only ciliary movements are observable in anoxia and severe hypoxia. Elevated DHE fluorescence indicated superoxide formation in near-anoxia, which might be attributable to the diminished SOD activity at low tissue oxygenation. However, we suggest that M. lignano could also be an interesting model to study hypothetical respiratory chain superoxide formation at low tissue oxygenation under the possible influence of hypoxic NO formation. The response to hyperoxia also differed from that of mammalian systems, but aligned with other marine invertebrate infauna that increase respiration rates in a P_O2-dependent manner above an upper critical P_c2. This is often interpreted as an attempt to reduce excess oxygen in the animals' environment and could be characteristic of species that have successfully adapted to deal with the vagaries of environmental oxygenation in intertidal sediments.

The authors would like to thank Nelly Tremblay for her advice and comments on the respiration measurements and further data analysis, Stefanie Meyer for her valuable laboratory assistance and Peter Karuso, Achim Grube and Matthias Köck for kindly supplying the Ageladine-A used in the present study. Thanks also go to Dr V. Lushchak and one anonymous referee for the comments on the original manuscript.

 
• C-H2DFFDA

  5-(and-6)-carboxy-2′,7 h-difluorodihydrofluorescein diacetate

 
• DHE

  dihydroethidium

 
• NH-FSW

  filtered sea water supplemented with 15 mmol l⁻¹ Na-Hepes

 
• P_c

  critical oxygen pressure

 
• pH_i

  intracellular pH

 
• P_O2

  oxygen partial pressure

 
• ROS

  reactive oxygen species

 
• SDT

  sodium diethyl-dithiocarbamate trihydrate

 
• SOD

  superoxide dismutase

 
• Δψ_m

  mitochondrial membrane potential