Oxidative gating of water channels (aquaporins) in Chara by hydroxyl radicals


Ernst Steudle. Fax: + 49 921 55 2564; e-mail: ernst.steudle@uni-bayreuth.de


Hydroxyl radicals (*OH) as produced in the Fenton reaction (Fe2+ + H2O2 = Fe3+ + OH + *OH) have been used to reversibly inhibit aquaporins in the plasma membrane of internodes of Chara corallina. Compared to conventional agents such as HgCl2, *OH proved to be more effective in blocking water channels and was less toxic to the cell. When internodes were treated for 30 min, cell hydraulic conductivity (Lp) decreased by 90% or even more. This effect was reversed within a few minutes after removing the radicals from the medium. In contrast to HgCl2, radical treatment reduced membrane permeability of small lipophilic organic solutes (ethanol, acetone, 1-propanol, and 2-propanol) by only 24 to 52%, indicating some continued limited movement of these solutes across aquaporins. The biggest effect of *OH treatment on solute permeability was found for isotopic water (HDO), which largely used water channels to cross the membrane. Inhibition of aquaporins reduced the diffusional water permeability (Pd) by about 70%. For the organic test solutes, which mainly use the bilayer to cross the membrane, channel closure caused anomalous (negative) osmosis; that is, cells had negative reflection coefficients (σs) and were transiently swelling in a hypertonic medium. From the ratio of bulk (Lp or osmotic permeability coefficient, Pf) to diffusional (Pd) permeability of water, the number (N) of water molecules that align in water channels was estimated to be N = Pf/Pd = 46 (on average). Radical treatment decreased N from 46 to 11, a value still larger than unity, which would be expected for a membrane lacking pores. The gating of aquaporins by *OH radicals is discussed in terms of a direct action of the radicals when passing the pores or by an indirect action via the bilayer. The rapid recovery of inhibited channels may indicate an easy access of cytoplasmic antioxidants to closed water channels. As hydrogen peroxide is a major signalling substance during different biotic and abiotic stresses, the reversible closure of water channels by *OH (as produced from H2O2 in the apoplast in the presence of transition metals such as Fe2+ or Cu+) may be downstream of the H2O2 signalling. This may provide appropriate adjustments in water relations (hydraulic conductivity), and a common response to different kinds of stresses.


Water channels (aquaporins) play an important role in water relations. In plants, 75 to 95% of trans-membrane movement of water is across these proteins (Steudle & Henzler 1995; Maurel 1997; Kjellbom et al. 1999; Tyerman et al. 1999; Steudle 2001; Maurel & Chrispeels 2001). Current research on plant water channels is focused on the gating of aquaporins, namely on opening or closing mechanisms which are thought to play a key role in the adaptation of plants to different kinds of stresses and, perhaps, in the cross-linking of these stresses (Steudle 2000; Javot & Maurel 2002; Pastori & Foyer 2002; Tyerman, Niemietz & Bramley 2002). Numerous internal or external factors control the gating of water channels such as their phosphorylation, the action of the water stress hormone ABA, pH, pCa, osmotic pressure and salinity, heavy metals, and temperature (Steudle & Tyerman 1983; Azaizeh, Gunse & Steudle 1992; Henzler & Steudle 1995; Johannson et al. 1996; Hertel & Steudle 1997; Zhang & Tyerman 1999; Hose, Steudle & Hartung 2000; Gerbeau et al. 2002; Niemietz & Tyerman 2002; Lee et al. 2003; Wan, Steudle & Hartung 2004; Ye, Wiera & Steudle 2004). Low nutrients, drought, anoxia, time of day or root development may be factors as well (Zhang & Tyerman 1991; Henzler et al. 1999; Wan & Zwiazek 1999; Clarkson et al. 2000; Barrowclough, Peterson & Steudle 2000; Tsuda & Tyree 2000; Martre, North & Nobel 2001; Hukin et al. 2002). Recently, a mechanical gating of water channels has been proposed. In root cortical cells, Wan et al. (2004) initiated channel closure by pulses of turgor pressure which induced a proportional diminution in water flow. The pulses resulted in an injection of kinetic energy to the channel protein causing a conformational change and its closure. A different mechanism was proposed by Ye et al. (2004). These authors induced substantial tensions (negative pressures) in the water-filled pores (aquaporins) of Chara, which then collapsed causing a reversible interruption of water flow and a decrease in water permeability. The cohesion/tension mechanism may explain the finding that, at least in some species, water permeability is reduced in the presence of high external solute concentration, such as salinity (Steudle & Tyerman 1983; Azaizeh et al. 1992). Mechanical gating may play a role during the adjustment of water permeability of living tissue surrounding xylem vessels that are under tension in transpiring plants. Such a mechanism has been proposed for roots (Passioura 1988; Kramer & Boyer 1995; Steudle & Peterson 1998). It has been known for a long time that root hydraulic conductivity (Lpr) increases as the tension in the xylem increases due to a high demand for water by the shoot.

Herein we describe a new type of an effective and reversible gating of water channels by hydroxyl radicals (*OH), which may be called an ‘oxidative gating’. Oxidative gating may be involved in ‘cross-talk’ between different stresses such as cold, drought and high light (Xiong, Schumaker & Zhu 2002). It is also known that reactive oxygen species (ROS) such as superoxide anion radical, H2O2, and *OH act as stressors (Pastori & Foyer 2002). During an oxidative burst, they are produced to protect plants against an invasion by pathogens (Wojtaszek 1997). ROS play a role as signalling agents as well (Pei et al. 2000). For the latter function, most are known for hydrogen peroxide and nitric oxide (Neill, Desikan & Hancock 2002a; Neill et al. 2002b). Recently, it has been shown that the green alga Chara corallina can tolerate levels of H2O2 of as high as 350 m m (Henzler & Steudle 2000). However, it turned out that Chara is highly sensitive to very low concentrations of hydroxyl radicals and that *OH substantially affects the water permeability of its membranes. Hence, there may be an inter-relation between oxidative stress (redox state) and water relations, and this may hold for other species as well. To demonstrate the action of the highly reactive oxidant *OH, hydroxyl radicals were produced in this research by the Fenton reaction (Fe2+ + H2O2 = Fe3+ + OH + *OH) outside of isolated Chara internodes. In the presence of Fe2+, internodes could tolerate H2O2 only at concentrations of up to a fraction of a millimole. With the aid of a cell pressure probe, the effects of *OH on the water permeability (hydraulic conductivity, Lp) were measured as well as those on the permeability (Ps) and reflection coefficients (σs) of certain solutes which moved across the membrane rapidly (acetone, monohydric alcohols, heavy water). As with mercurials (Henzler & Steudle 1995), *OH induced anomalous (negative) osmosis in the presence of the test solutes. The effects were more pronounced than with the conventional inhibitor mercuric chloride (HgCl2). Among the reactive oxidative species such as hydrogen peroxide or superoxide anion radical, the hydroxyl radical is the most reactive compound. The molecule contains an unpaired electron and tends to rapidly attack and oxidize neighbouring molecules, namely organic compounds. When *OH takes an electron from another non-radical, that molecule becomes a radical. This may initiate a chain reaction of electron removal that will eventually result in destruction of bioactive macromolecules such as proteins, sugars, membrane lipids, nucleotides, and organic materials (Tien, Svingen & Aust 1982; Stadtman 1993; Samaha et al. 1999; Nakamura, La & Swenberg 2000). The action of *OH on water channels could have been either directly from inside the pore, when the molecule moves across, or indirectly by lipid peroxidation and an attack by resulting radicals from outside. The action of *OH radicals on water channels of Chara was largely reversible. As with the precise mechanism by which *OH acts on aquaporins, the mechanism of the recovery of aquaporin activity is not yet clear. Different mechanisms are discussed such as an action of cytoplasmic antioxidants (e.g. ascorbate) directly on the aquaporin or indirectly on the bilayer which has been subject to peroxidation in the presence of the oxidant.


Plant material

Chara corallina was grown in artificial pond water (APW; composition in mole m−3: 1.0 NaCl, 0.1 KCl, 0.1 CaCl2 and 0.1 MgCl2) as described previously (Henzler & Steudle 1995) in tanks that contained a layer of natural pond mud. The plants were continuously illuminated with a 15 W fluorescent lamp (Herman-Electronic, Hohentengen, Germany) positioned 0.2 m over the water surface. Chara internodes freed from adjacent cells were incubated in APW for several hours before the experiments were started. The internodes used in experiments were 50 to 150 mm in length and 0.8 to 1.0 mm in diameter.

Determination of transport parameters (Lp, Ps and σs)

There are three important transport parameters that can be measured by cell pressure probes (Steudle 1993). The hydraulic conductivity (Lp) is a measure of water permeability, the solute permeability coefficient (Ps) denotes the passive permeability of the cell membrane for a given solute, and the reflection coefficient (σs) is a quantitative measure of the ‘passive selectivity’ of the cell membrane for that solute as compared to water. Usually, plant cell membranes would have a σs between zero and unity. When σs = 0 the membrane does not distinguish between solute and water. Both pass at the same rate. When σs = 1 the membrane is ideally semi-permeable. In this case, only water can pass through the membrane and Ps = 0. The respective solutes are completely ‘reflected’ by the membrane. There are also cases in which σs < 0 (Henzler & Steudle 1995). This refers to the striking situation that a cell does not shrink but swells in hypertonic media because, during osmosis, solutes are getting into the cell faster than the water can get out. This phenomenon, which can be observed with plant cells, is called ‘anomalous (negative) osmosis’. Equations for calculating Lp, Ps and σs are:


where V is the cell volume; A is the cell surface area; πi is the osmotic pressure of cell sap; ɛ is the elastic coefficient of the cell (elastic modulus); ks is the rate constant of solute exchange; Po − Pmin(max) is the maximum change in cell turgor pressure; RT · inline image was the given change of osmotic pressure of the medium. For a detailed description of these equations, the reader is referred to earlier publications such as Steudle (1993), Henzler & Steudle (1995) or Ye et al. (2004).

Effects of *OH on the transport of water and solutes

As described earlier, the cell pressure probe (completely filled with silicone oil) was introduced through the protruding node adjacent to a Chara internode which had been placed in a glass tube which had an inner diameter of 3 mm and fixed by a clamp (Henzler & Steudle 1995; Hertel & Steudle 1997).  APW or test solutions were pumped through the other end of the glass tube along the cell (flow rates were 0.15–0.2 m s−1), so that the solution around the cell was vigorously stirred and exchanged within a fraction of a second. This minimized the thickness of external unstirred layers (Steudle & Tyerman 1983). In order to induce pressure relaxations, the oil/cell sap meniscus forming in the tip of the capillary was moved forward or backward and was kept stable after each move. From the half time (inline image) of pressure relaxations, Lp was calculated as a control (see Eqn 1). Solutes for testing the permeability of the membrane were heavy water (HDO at 3800–4750 m m; Merck, Darmstadt, Germany), acetone (134–173 m m), ethanol (110–140 m m), 1-propanol (155–173 m m) and 2-propanol (144–166 m m). Control values of Ps and σs were calculated from biphasic response curves (Steudle & Tyerman 1983; Steudle 1993; Eqns 2 and 3). Hydroxyl radicals were produced by the Fenton reaction:

Fe2 +  + H2O2 = Fe3 +  + OH −  + *OH (4)

in the presence of 0.6 m m H2O2 and 3 m m FeSO4 added to APW. This should have produced *OH at a very low concentration (see Discussion). Following the addition of H2O2, changes in half times of water flow (inline image) were measured during pressure relaxations which were produced every 2 min using the probe. After about 30 min, a steady half time of pressure relaxation (inline image) was attained. After reaching the maximum inline image (minimum Lp), test solutes (HDO, acetone, ethanol, 1-propanol and 2-propanol) were added to the medium to measure changes of their Ps and σs in response to the radical treatment. For acetone, however, measurements of Ps and σs were also performed for some cells during the period of increasing inline image.

To remove *OH, the internodes were rinsed with control medium (APW). Again, using the cell pressure probe, inline image was checked every 2 min to test for the reversibility of the effects of *OH. When control values of inline image were re-attained, the same test solutes (HDO, acetone, ethanol, 1-propanol and 2-propanol) were used to re-examine the reversal of their effects. For a given cell, the whole time course of the experiments lasted for 2–4 h. During treatments, cells did not lose turgor pressure (0.6–0.7 MPa) within ± 0.03 MPa or ± 5%.


In the presence of *OH, the activity of aquaporins was dramatically inhibited. Figure 1a shows the time course of changes in inline image (∼1/Lp) during radical treatment for a given Chara internode. The value of inline image increased continuously during treatment and attained a constant value after about 30 min. In the example shown in Fig. 1a, inline image increased by a factor of 15. Hence, Lp was reduced by 93% in this case. Reductions in Lp varied from 90 to 95% (n = 15 cells). This value was significantly larger than that measured earlier in the presence of 50 µm HgCl2 which only reduced Lp by 75% in Chara (Henzler & Steudle 1995). As inline image increased, there was also an increase of the half time of permeation of the test solute acetone (inline image; Fig. 1b). The value of inline image was approximately doubled, which resulted in a reduction of the solute permeability by a factor of two. Even though this was much less than the reduction in Lp, it was still substantial. In earlier experiments with HgCl2, there was no significant change in Ps for lipophilic solutes such as acetone (Table 1; Steudle & Henzler 1995). This had been interpreted by assuming a different pathway for the solutes, i.e. preferentially across the bilayer (Steudle & Henzler 1995; Hertel & Steudle 1997). The present results show that this conclusion has to be modified and that there is some movement of small uncharged solutes across water channels in Chara (see Discussion). As shown in Fig. 1, removal of *OH by washing the cell with control medium (APW), resulted in re-attaining both the original inline image and inline image to a good approximation (90% recovery for water and 92% for acetone).

Figure 1.

Increase of half times of both water exchange (inline image; a) and acetone permeation (inline image; b) in an internode of Chara (inline image ∼1/Lp; inline image ∼ 1/Ps) caused by a treatment with *OH radicals. Closure of water channels was complete after about 30 min of treatment. An increase in inline image by a factor of 15 indicates a decrease of Lp by the same factor. inline image increased by a factor of two indicating that some of the solute did use aquaporins to cross the membrane. After removal of *OH radicals from the medium, the original inline image was re-attained within a few minutes indicating a fast recovery of water channel activity.

Table 1.  Effects of *OH radical treatment on half times of water exchange inline image of Chara internodes, and on permeability (Ps) and reflection (σs) coefficients of test solutes (heavy water HDO, acetone and ethanol)
TreatmentHalf time of water exchange, inline image (s)Test
Permeability coefficient, Ps (10−6 m s−1)Reflection coefficient, σs (1)
  1. Values are means ± SD; n = 5–15 cells; C, control; T, *OH treatment; R, recovery; T/C, relative change caused by treatment. Comparison with results from an earlier study by Henzler & Steudle (1995) using the channel blocker HgCl2 at a concentration of 50 µm indicates that *OH radicals were more effective in blocking water channels in Chara. T/C ratios are given bold to underline differences between HgCl2 and *OH Freatments. For further explanation, see text.

HgCl2 (50 µm)3.6 ± 1.012.3 ± 3.83.8 ± 1.0 3.4HDO7.7 ± 3.04.3 ± 2.00.560.0034 ± 0.00070.0070 ± 0.00090.0039 ± 0.00062.1
acetone4.1 ± 0.23.7 ± 0.20.900.13 ± 0.04−0.16 ± 0.04
ethanol2.7 ± 1.02.9 ± ± 0.06−0.07 ± 0.07
*OH2.0 ± 0.829.3 ± 8.62.4 ± 1.114.5HDO6.9 ± 0.81.9 ± 0.76.5 ± ± 0.00010.0015 ± 0.00060.0006 ± 0.00013.0
acetone3.8 ± 0.11.8 ± 0.33.3 ± 0.20.470.15 ± 0.02−0.43 ± 0.110.13 ± 0.03
ethanol3.5 ± 0.31.9 ± 0.53.2 ± 0.30.540.21 ± 0.02−0.25 ± 0.090.19 ± 0.09

In Fig. 2, typical relaxation curves (hydrostatic and osmotic) in response to *OH treatment are shown (steady state), again using acetone as the permeating test solute. Hydrostatic pressure relaxations are given on the left side of the figure. Usually, in the presence of a permeating solute such as acetone, osmotic response curves were biphasic (right side of the figure; Steudle & Tyerman 1983; Steudle 1993). There was an initial phase during which turgor pressure rapidly decreased or increased due to an exosmotic or an endosmotic water flow, respectively. This so-called ‘water phase’ was rapid because of the high permeability of the cell membrane to water. It was followed by a ‘solute phase’. During this second phase, turgor increased or decreased again due to the passive flow of solute into or out of the cell tending to equilibrate the concentration of permeating solutes on both sides of the membrane, and water followed the movement of the solute (Fig. 2a). As shown in Fig. 1a, blockage of water channels with *OH increased inline image by a factor of 10 or even more (Fig. 2b). Most interestingly, upon water channel closure, the response to hypertonic solution was a transient increase in turgor (anomalous osmosis). The permeating solute (acetone) entered the cell faster than the water could get out. As a consequence, the cell did not shrink (turgor pressure decrease) but was swelling (turgor pressure increase) in hypertonic solution. The phenomenon of anomalous (negative) osmosis is described by a negative reflection coefficient. For the example given in Fig. 2b, σs decreased to −0.46. Again, the figure shows a two-fold inhibition of solute permeability. When *OH was removed from the medium, the inhibition of water and solute flow was reversed within a short period of time, inline image recovered to 90% of the original value, and the solute parameters (Ps and σs) recovered by 96 and 85%, respectively (Fig. 2c).

Figure 2.

Effects of *OH treatment on hydrostatic and osmotic pressure relaxations as measured with a pressure probe in the steady state. Compared with the control, hydrostatic half times increased by a factor of 10. With the permeating solute acetone, responses were biphasic. The reflection coefficient in the control was smaller than unity (σs = 0.13). Water channel closure resulted in anomalous osmosis (negative reflection coefficient, σs = −0.46). Because the solute entered the cell faster than the water could get out, the internode was swelling in a hypertonic solution. From the second phase of responses, permeability coefficients of acetone were calculated (Eqn 2). Treatment affected solute transport much less than that of the water. A substantial amount of solute was permeating across water channels which were not completely selective for water. Removal of *OH from the solution resulted in a recovery of permeabilities for water and solute and of reflection coefficients by 90% (Lp), 96% (Ps) and 85% (σs) of the original values.

In the presence of hydrostatic or osmotic gradients of pressure, there should be a bulk water flow across aquaporins that should differ from the diffusional water flow through aquaporins as measured with isotopic water (heavy water, HDO). The ratio between the bulk (Pf; see Eqn 5) and diffusional (Pd) water permeability is a measure of the number of water molecules within the pores (Levitt 1974; see Discussion). When closing aquaporins by radical treatment, we should thus expect to see big changes in the Pf/Pd ratio as most of the water is now using the bilayer to cross the membrane. Diffusional water flow across the bilayer should result in a Pf/Pd ratio of unity. Measured changes of Lp (Pf) and Pd values are given in Fig. 3 (mean ± SD; n = 5–15 cells). Following the inhibition of water channels by *OH, cell Lp decreased to about 1/10 of the control value indicating a closure of most of the channels. The osmotic permeability Pf shown in Fig. 3b is just proportional to Lp according to the relation:

Figure 3.

Summary of the effects of *OH radical treatment on Lp and Pf (= Lp · RT/inline image), and on the diffusional water permeability, Pd, as measured with heavy water (mean values ± SD; n = 5–15 cells). (a) Closure of water channels by *OH radicals reduced Lp (Pf) to 10% of control. After removing *OH radicals from the medium, Lp (Pf) recovered to 85% of its original value. (b) Comparison between Pf and Pd shows that Pf >> Pd. Inhibition of aquaporins reduced the diffusive permeability of isotopic water (HDO) to 30% of the control. This effect was reversed to 95% of the original when *OH was removed. (c) From the ratio of bulk to diffusive permeability of water (Pf/Pd), the number (N) of the water molecules that align in water channels was estimated to be N = 35–60. In the presence of closed channels, N decreased to 11.


Reductions of Lp and Pf were as large as 90% of the control (Fig. 3a). However, the reduction in Pd was 70% as compared with the control, namely smaller (Fig. 3b). The average value of the Pf/Pd ratio was 46 in the control and 11 following the *OH treatment. The reduction of the Pf/Pd ratio was significant (P < 0.05). *OH can be removed simply by washing Chara internodes with control medium APW. After the removal of *OH, water permeability Lp and Pf recovered to 85% of the controls within a few minutes (see also Fig. 1) indicating that water channels re-opened rather quickly due to the action of repair mechanisms of the cell. To a large extent, the ratios of Pf/Pd recovered as well. The results indicate that *OH treatment caused a reduction of the porous channels as compared with the bilayer pathway, although ratios were still substantially bigger than unity after channel closure (see Discussion).

Isotopic water (HDO) should be a good tracer for normal water. It should mainly use aquaporins to diffuse across cell membranes as verified in Fig. 3b. Hence, the effect of channel closure on Pd was the largest for this solute (reduction by 70%). For other test solutes, reduction of Ps ranged between 24 and 52% (acetone, 52%; ethanol, 46%; 1-propanol, 37%; 2-propanol 24%). The difference between HDO and the other solutes is evident from the fact that channel closure caused a decrease of the reflection coefficients of the latter even to negative values down to –0.50. However, the σs value of HDO increased, although absolute values were close to zero (Fig. 4). In terms of the composite transport model (Steudle & Henzler 1995), the result is understandable. Upon complete channel closure, reflection coefficients should assume values of the bilayer, which should be quite low or even negative for lipophilic solutes. However, for HDO, reflection coefficients of the bilayer should be bigger than those of water channels (Henzler & Steudle 1995). Hence, the σs of HDO remained positive and increased by a factor of three. As observed for water permeability, the permeability and reflection coefficients of test solutes tended to not completely recover after removal of *OH. However, original and recovered values were not significantly different (P > 0.05).

Figure 4.

Summary of diffusional permeabilities (Ps) and of reflection coefficients (σs) for some small, lipophilic organic solutes and heavy water as affected by *OH treatment (mean values ± SD; n = 5–6 cells). To cross the membrane, organic solutes should largely use the bilayer. Heavy water should use aquaporins. Upon channel closure, there was a reduction of solute permeability which was the biggest for heavy water. The reflection coefficient of this solute, however, increased upon channel closure and did not decrease as was the case for the other solutes (acetone, ethanol, 1-propanol, and 2-propanol). Note that because of the small absolute value of the reflection coefficient of HDO, this is given as a multiple of 10. Effects nearly recovered after removal of *OH radicals.


For the first time, hydroxyl radicals (*OH) have been used to reversibly inhibit the activity of water channels in internodes of Chara by an ‘oxidative gating’. The effect was as large as one order of magnitude for water, and was much bigger than that of conventional aquaporin inhibitors such as HgCl2. Mercuric chloride has the disadvantage that it is quite toxic and inhibits many cell functions. When keeping its concentration low, it appeared that toxic effects of *OH were less pronounced than those of HgCl2. It should be noted that the actual concentration of *OH generated in the present study must have been quite low. In the presence of a fraction of a millimole of H2O2, the actual concentration of *OH should have been in the nanomolar range due to its extremely short half-life (10−9 s; Caro & Puntarulo 1996; Agarwal, Saleh & Bedaiwy 2003). In the absence of Fe2+, Chara internodes can tolerate H2O2 in concentrations of up to 350 m m without affecting turgor and water transport across cell membranes (Henzler & Steudle 2000). We verified that 3 m m FeSO4 by itself did not affect inline image and inline image even during long treatments. Hence, the effective agent was *OH. Cell turgor pressure was between 0.6 and 0.7 MPa. Turgor remained constant within ± 0.03 MPa during the experiments, indicating that the integrity of cell membranes was maintained.

The mechanism by which *OH acts on aquaporins is not yet understood. The highly reactive nature of *OH stems from its unpaired valence electron that causes high oxidative reactivity. *OH attacks and damages any organic matter such as nucleic acids, membrane lipids, proteins, nucleotides, and carbohydrates. The rates at which *OH reacts with organic material are extremely high. Rate constants of 107−1011 m−1s−1 have been reported with almost every type of molecule found in living cells (Halliwell & Gutteridge 1989). Due to its high reactivity, it is unlikely that *OH produced in the medium moved across the Chara cell wall of 5–10 µm without being completely absorbed in reactions with wall material such as during the oxidative scission of plant cell wall polysaccharides (Fry 1998). It is more likely that *OH radicals were produced close to the plasma membrane where Fe2+ should have been located due to negative fixed charges in the wall. Since *OH does not bear charges and is even smaller than water, it should be able to pass rapidly through the aquaporins right after its generation close to the membrane. Within the channel, it may attack cysteine or other residues as shown in the literature (Preston et al. 1993; Sanchez-Gongora et al. 1997; Caselli et al. 1998; Soto et al. 2002). Oxidation of aquaporins, in turn, may cause a change in the conformation of the protein and result in channel closure. In an alternative scenario, *OH may oxidize lipids by attacking C = C bonds and, thus, change the milieu around water channels (e.g. its polarity). Attack of the protein by lipid radicals, in turn, could result in a channel closure. To date, we can only speculate about possible mechanisms. There is no evidence regarding which of the proposed mechanisms is more likely. In the experiments it took about 30 min to reach the maximum effect of inhibition of channels. This could be explained by either mechanism, and a dynamic balance between damage and repair by antioxidants produced in the cytoplasm.

Plants possess very efficient scavenging antioxdant systems that protect them from destructive oxidative reactions. The fact that inhibition of water channels was reversed within a few minutes after removal of *OH from the medium indicated an effective repair mechanism(s), which should be related to cytoplasmic antioxidants such as ascorbate, glutathione or NADPH. Ascorbate usually acts as the reducing component which is then scavenged by glutathione and NADPH (Foyer & Lelandais 1993; Foyer, Descourvières & Kunert 1994; Schützendübel & Polle 2002). When channel closure is caused by lipid oxidation, recovery would be due to regeneration of the lipids around the channels and regeneration of oxidized protein from ‘outside’. On the other hand, when the channel protein is oxidized from ‘inside’, the big scavenger molecules may have problems to act on the inside of the pore. However, the channel inside may become accessible during the conformational change, thereby exposing oxidized parts to the cytoplasm.

Reversible closure of aquaporins by nanomolar concentrations of *OH in Chara are similar to those obtained in the presence of the conventional water channel blocker HgCl2 (Chrispeels & Maurel 1994; Henzler & Steudle 1995; Tazawa, Asai & Iwasaki 1996). When Chara internodes were treated with 50 µm HgCl2 for 10–15 min, cell Lp was reduced by 75%. It recovered in the presence of 5 m m of the scavenger 2-mercaptoethanol (Henzler & Steudle 1995), which removed the mercury from SH groups of the protein. However, HgCl2 caused side-effects when applied for longer periods of time. As a consequence, there was irreversible damage evidenced by a continuous decline in cell turgor pressure even when HgCl2 was removed immediately after the experiment (Steudle & Henzler 1995; Zhang & Tyerman 1999). It was not possible to use concentrations of HgCl2 higher than 50 µm in order to produce reductions of larger than 75%. Although there was not a complete recovery after *OH treatment (Fig. 3a), the present results indicate that *OH is much more effective and less toxic than HgCl2 in blocking water channels in Chara. In longer terms, there may be a complete recovery of water and solute permeability from *OH treatment, an idea that is currently being tested. The fact that more water channels could be closed in the presence of *OH than by using HgCl2 may be due to the existence of a population of different water channels instead of just one type. Some of the channel proteins may have SH groups and could be attacked by HgCl2; some of them do not have such groups or they are not accessible by mercurials, and are not sensitive to HgCl2 (Daniels, Mirkov & Chrispeels 1994; Biela et al. 1999; Krajinski et al. 2000). However, these aquaporins may be accessible to *OH, and more types of water channels may be affected by *OH. This idea is consistent with results of molecular studies showing that plants have quite a number of putative aquaporins (Weig, Deswarte & Chrispeels 1997; Chaumont et al. 2001).

Unlike HgCl2, water channel closure by *OH also affected transport of small organic solutes such as acetone. Solute permeability was also substantially reduced for monohydric alcohols. As one would expect, the effect was biggest for heavy water (HDO). This solute should largely use water channels to cross the membrane (Ye et al. 2004). The effect of *OH on solute permeability supports the earlier view of a limited passage of small organic molecules across water channels in Chara (Steudle & Henzler 1995; Henzler & Steudle 1995; Hertel & Steudle 1997). It is in line with results from other research on aquaporin selectivity indicating that aquaporins may allow the passage of substances such as glycerol, urea, amino acid or even to small peptides (Kammerloher et al. 1994; Rivers et al. 1997; Agre, Bonhivers & Borgnia 1998; Borgnia et al. 1999; Dean et al. 1999; Gerbeau et al. 1999).

The movement of small organic solutes through water channels in Chara should have consequences for the overall osmotic properties of cell membranes and also their selectivity as expressed by the reflection coefficient (σs). A composite transport model of the plasma membrane has been used to explain absolute overall values of reflection coefficients and how they would change upon channel closure (Henzler & Steudle 1995). According to the model, there are two different arrays (water channels and lipid bilayer) in the cell membranes which contribute to the overall reflection coefficient of a given solute. The σs measured with a cell pressure probe in present experiments would, thus, be a composite of water channel and lipid bilayer arrays. For HDO that should largely use the water channel path, σs increased by a factor of three upon the closure of water channels by *OH as expected from the model. The σs of the lipid bilayer should be small or even negative for rapidly permeating solutes such as acetone, ethanol, 1-propanol, 2-propanol and so on, which mainly diffuse across cell membranes through the bilayer. In this case, water channel closure should result in a decrease of σs (even to negative), as found. Changes of σs for HDO (increase), acetone and ethanol (decrease to negative) were more pronounced when water channels were blocked by *OH than with HgCl2 (Table 1). The effect on σs can be dramatic. It is in line with the composite transport model. This is clear evidence for the role of water channels and their ability to allow the passage of uncharged small solutes in addition to water. As expected, passage was the biggest for HDO, but the contribution to the other solutes can be substantial as well depending on how effective the solute was fitting into the pore. Size and polarity of test solutes should have been important, and their ability to form hydrogen bonds within the pore.

At the present, we are not in a position to calculate the precise number of water molecules in the pore of a Chara aquaporin. According to Levitt's (1974) theory, the ratio between bulk or osmotic (Pf ∼ Lp; Eqn 5) and diffusional (Pd) water flow directly yields the number (N) of water molecules aligned in water channels; namely


This assumes that water transport through water channels can be identified with the overall hydraulic conductivity (Lp), whereas Pd can be identified with the diffusional water permeability of channels. Present results shown in Fig. 3c suggest that there are 46 (on average) water molecules aligned in a water channel of Chara, which is larger than estimates from previous results (31 molecules/channel, Hertel & Steudle 1997; 27 molecules/channel, Henzler & Steudle 1995). It is also larger than the figure reported for red blood cells (10 molecules/channel, Finkelstein 1987). In part, the difference between Chara and red blood cells had been explained by the existence of unstirred layers which may have a dramatic effect on Pd but not on Pf, resulting in an overestimation of N (Steudle & Tyerman 1983; Henzler & Steudle 1995; Hertel & Steudle 1997). Using the same set-up for the internodes in a perspex tube as herein, Steudle & Tyerman (1983) measured the response in rates of transport (water, solutes) to the rate of stirring of the external medium. They concluded that, at maximum, external unstirred layers could be as thick as 50 µm. In view of diffusional exchange rates of 20–50 s for heavy water and for the rapidly permeating solutes, there should have been no significant effect on Pd, Ps and σs by external unstirred layers (see Discussion in Henzler & Steudle 1995; and in Hertel & Steudle 1997). At a radius of internodes of approximately 0.5 mm, internal unstirred layers could be as large as a few hundred micrometers, which is not negligible. However, there was some stirring by protoplasmic streaming and a substantial reduction by the cylindrical geometry of cells so that effects should have been much smaller (see Discussion in Hertel & Steudle 1997). Solute phases of HDO and of the other rapidly permeating solutes were nicely exponential throughout. This indicated that the contribution of internal unstirred layers was rather low. If unstirred layers build up, this would increase inline image during the solute phase. Hence, effects on Pd should have been relatively small. However, the present data do not permit the quantification of the number of molecules aligned in a channel (although providing an upper limit). This would have been possible, if treatment by *OH would have closed all membrane pores which could allow a passage of water, that is all aquaporins as well as all other transporters such as pumps and ion channels. Then the assumption could be made that, after closure, Pf = Pd. Hence, it would be possible to work out Pf and Pd of the aquaporins themselves and its true N-value, as has been done for red blood cells (Mathai et al. 1996).

The physiological significance of the gating of water channels by *OH radicals needs to be addressed briefly. Hydrogen peroxide (and nitric oxide, NO) are known to be involved during signal transduction in response to many different stresses such as drought, high salinity, oxygen deprivation, chilling, and osmotic stress (Xiong et al. 2002). However, it is obvious, at least for Chara, that there is no direct effect on water transport of the systemic compound H2O2. Previous results indicated that the alga tolerates concentrations of H2O2 as high as 350 m m. Aquaporins in Chara were highly permeable to this solute, which has a chemical structure similar to that of water (Henzler & Steudle 1995, 2000). Indeed, aquaporins appeared to act also as ‘peroxoporins’. In the present paper, we show that there is a strong response to locally produced, short-lived and highly reactive hydroxyl radicals which are also thought to act during ROS signalling (Neill et al. 2002a, b). Hence, the apparent signalling of H2O2 in Chara is a consequence of its conversion into *OH radicals in the apoplast in the presence of a transition metal. This type of a downstream mechanism of the signalling of H2O2 may be important in higher plants. *OH may be produced in cell walls close to the membranes in response to the major systemic signal substance H2O2 in the presence of transition metal ions such as Fe2+ or Cu+ (Fry 1998). Experiments with higher plant tissues are underway to test this hypothesis. In a tissue apoplast, it is, however, more difficult than for an isolated cell to produce *OH radicals in a defined way close to plasma membranes.

In conclusion, the results show that the activity of water channels in the cell membranes of Chara corallina can be substantially and reversibly inhibited by *OH. Compared with conventional blockers of aquaporins such as mercurials, *OH turned out to be more effective in blocking aquaporins and less toxic for cells. Water permeability (Lp) was reduced by more than 90% when using *OH and recovered to 85% of the control when *OH was removed. Unlike HgCl2, *OH reduced the permeability of small uncharged solutes indicating some transport of these solutes across the pores in addition to water. For rapidly permeating lipophilic solutes, the blockage of water channels with *OH resulted in negative reflection coefficients and anomalous osmosis as expected from the composite transport model. From the ratio of bulk to diffusive permeability of water, the number of water molecules that line up in channels was estimated to be N = 46 on average. This figure may represent an overestimate due to effects of unstirred layers, and a number of some 20 molecules/channel is, perhaps, more realistic. Treatment with *OH reduced N substantially as expected from the model. At present, we can only speculate about the mechanisms by which *OH acts on water channels. Two alternatives may be possible. One is that aquaporins were oxidized by *OH attacking the channel from inside the pore to cause conformational changes of the proteins and its closure. The other alternative is that C = C double bonds of the plasma membrane were attacked by *OH, resulting in the formation of aggressive radicals which attacked aquaporins from outside. Regardless of which of the mechanisms will turn out to be true in the future, our results indicate a regulation of water channel activity by an oxidative signalling initiated in the presence of *OH. This may also exist in higher plants providing an interaction between the redox state (oxidative stress) and water relations (water stress).


Thanks go to Professor Carol Peterson (University of Waterloo, Canada) and Jan Muhr (University of Bayreuth) for carefully reading the manuscript and making helpful suggestions. We are indebted to Burkhard Stumpf (Department of Plant Ecology, University of Bayreuth) for his expert technical assistance.