Oxidative gating of water channels (aquaporins) in corn roots


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


An oxidative gating of water channels (aquaporins: AQPs) was observed in roots of corn seedlings as already found for the green alga Chara corallina. In the presence of 35 m m hydrogen peroxide (H2O2) – a precursor of hydroxyl radicals (*OH) – half times of water flow (as measured with the aid of pressure probes) increased at the level of both entire roots and individual cortical cells by factors of three and nine, respectively. This indicated decreases in the hydrostatic hydraulic conductivity of roots (Lphr) and of cells (Lph) by the same factors. Unlike other stresses, the plant hormone abscisic acid (ABA) had no ameliorative effect either on root Lphr or on cell Lph when AQPs were inhibited by oxidative stress. Closure of AQPs reduced the permeability of acetone by factors of two in roots and 1.5 in cells. This indicated that AQPs were not ideally selective for water but allowed the passage of the organic solute acetone. In the presence of H2O2, channel closure caused anomalous (negative) osmosis at both the root and the cell level. This was interpreted by the fact that in the case of the rapidly permeating solute acetone, channel closure caused the solute to move faster than the water and the reflection coefficient (σs) reversed its sign. When H2O2 was removed from the medium, the effects were reversible, again at both the root and the cell level. The results provide evidence of oxidative gating of AQPs, which leads on to inhibition of water uptake by the roots. Possible mechanisms of the oxidative gating of AQPs induced by H2O2 (*OH) are discussed.


The gating (opening and closing) of water channels (aquaporins: AQPs) is thought to be of key importance in the regulation of water transport across plant cell membranes. There are numerous internal or external factors that cause the gating of AQPs, such as high concentration or salinity, pH, pCa and heavy metals (Steudle & Tyerman 1983; Azaizeh, Gunse & Steudle 1992; Henzler & Steudle 1995; Gerbeau et al. 2002; Niemietz & Tyerman 2002; Ye, Wiera & Steudle 2004; Ye, Muhr & Steudle 2005). The activity of AQPs can also be affected by temperature, nutrient deprivation or hypoxia (Hertel & Steudle 1997; Zhang & Tyerman 1999; Clarkson et al. 2000; Lee, Chung & Steudle 2005a,b). It has been shown that protein phosphorylation can provide metabolic control of AQP activity (Johannson et al. 1996). The plant stress hormone abscisic acid (ABA) has an ameliorative effect that tends to either keep AQPs open or cause a rapid transition from the closed to the open state (Freundl, Steudle & Hartung 1998, 2000; Hose, Steudle & Hartung 2000; Wan, Steudle & Hartung 2004; Lee et al. 2005b). Other experiments have shown that there may be a diurnal rhythm in AQP activity that is related to diurnal synthesis and degradation of channel proteins (Henzler et al. 1999). Recent findings of Wan et al. (2004) indicated a mechanical inhibition of AQPs, which may be important during rapid changes in cell turgor as they occur during osmotic stress or rapid changes in transpiration. The list of factors may get longer in the future. In most cases, the precise mechanisms by which stresses or other factors affect the open/closed states of AQPs are not known.

Recently, Henzler, Ye & Steudle (2004) described a new type of oxidative gating of AQPs in the green alga Chara corallina. Hydroxyl radicals (*OH) produced during the Fenton reaction (Fe2+ + H2O2 = Fe3+ + OH + *OH) were used to inhibit AQP activity in the plasma membrane of Chara internodes. When cells were treated with *OH radicals for about 0.5 h, cell hydraulic conductivity (Lp) decreased by 90% or even more. The effect was reversed within a few minutes after removal of the radicals from the medium. Aroca et al. (2005) found that treatment with 100 µm hydrogen peroxide (H2O2) decreased the root hydraulic conductance of a chilling-sensitive maize genotype but had no effect on a chilling-tolerant genotype. These authors referred the changes to a membrane damage caused by H2O2 accumulation during chilling treatment to the chilling-sensitive genotype which did not recover upon the increase of AQP abundance and activity. Because H2O2 is a major signalling substance during different biotic and abiotic stresses (Pastori & Foyer 2002; Xiong, Schumaker & Zhu 2002), Henzler et al. (2004) speculated that 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+ (Fry 1998), may be a downstream reaction during H2O2 signalling. It may provide appropriate adjustments in water relations and a common response to different types of stresses from which plants may suffer.

In the present paper, the oxidative gating of AQPs is investigated further in experiments with young roots of corn (Zea mays L.). Because the nutrient solution already contained FeNaEDTA as Fe3+, Fe2+ should also be present as a result of the reaction with the superoxide anion (O2), such as Fe3+ + inline image = Fe2+ + O2 or other reactions either within cells or in the root apoplast (Chen & Schopfer 1999; Liszkay, Zalm & Schopfer 2004). Therefore, we used H2O2 at a concentration of 35 m m instead of a mixture of H2O2 and Fe2+ to produce *OH radicals [as in the Chara experiments of Henzler et al. (2004), where a mixture of 3 m m Fe2+ and 0.6 m m H2O2 was used]. The relatively high concentrations of H2O2 were not harmful to the roots, which reversibly tolerated them when applied for 2–3 h. Pressure probes were used to measure the effects on cell (in the outer cortex) and entire root hydraulic conductivity. Anomalous osmosis could be reversibly induced in the presence of the lipophilic solute acetone at both the root and the cell level. The phenomena have been interpreted in terms of the composite structure of the cell membrane and roots (Steudle & Henzler 1995; Steudle & Peterson 1998). The stress hormone ABA had an ameliorative effect during the inhibition of radial water permeability of roots (Lpr) and cell Lp by treatments like mechanical stimuli or low temperature (Freundl et al. 2000; Hose et al. 2000; Wan et al. 2004; Lee et al. 2005b). In the present paper, this effect was tested as well for the oxidative gating of AQPs.


Plant material

Seeds of corn (Zea mays L. cv. Helix; Kleinwanzlebener Saatzucht AG, Einbeck, Germany) were germinated on filter paper soaked with 0.5 m m CaSO4 for 3 d at 25 °C in the dark. When seminal roots were 30–50 mm long, seedlings were transferred to 7 L containers, which accommodated 20 seedlings each. For detailed information of growing conditions, the reader is referred to earlier publications such as Freundl et al. (1998), Hose et al. (2000) and Wan et al. (2004). Roots of 8- to 10-d-old seedlings (including the time required for germination) were used in the experiments. They measured 250–400 mm long.

Root pressure probe experiments

The end segments of the roots were placed in a glass tube (inner diameter: 6 mm) with the basal cut ends protruding and connected to a root pressure probe. Segments (length: 80–120 mm) were fixed to the probe by silicone seals (Zhu & Steudle 1991). Root medium flowed at a rate of 0.30–0.50 m s−1 along the roots by gravity from a reservoir sitting about 0.5 m above the glass tube (0.2 m long) and was circulated back to the reservoir by a peristaltic pump. The flexible silicone rubber tubes connecting the reservoir and the glass tube also had an inner diameter of 6 mm. Even though flow was turbulent, this should have resulted in an average flow rate of a few m s−1. This rate was much too high. It would have caused leakages in the roots or would have even broken them. Hence, the speed was reduced by using a two-way stopcock between the reservoir and the glass tube, which also allowed the changing of solutions using two different reservoirs (see fig. 1 in Azaizeh & Steudle 1991; Hertel & Steudle 1997). Average rates of water flow were regulated to 0.30–0.50 m s−1, which did not cause too much shaking of the roots or even damage but reduced unstirred layers to a minimum (Steudle & Tyerman 1983; Henzler & Steudle 2000). Water flow within the rubber tubing and the glass pipe was turbulent as the water stream was injected into the tubing through borings in the stopcock, which had a diameter of 2 mm and were arranged at an angle of about 30°.

Turbulences within the system could be seen from the movement of little air bubbles that occasionally passed through when changing solutions. In the tests, the turbulences were made visible by using finely suspended matter (murky solution obtained by shaking sand for gardening with the root medium). The turbulences were largely a result of changes in diameter and of flow being forced to deviate from a straight direction. In addition, the Reynolds number (Re) was close to the critical value of 2000, where laminar flow tends to become turbulent (Re = 1800–3000). During the travel of water along the tubing (0.5 m), the turbulences should have tended to even out. However, they were reinforced by a constriction (3 mm in diameter) at the entrance to the glass tube, causing an increase in the rate of the water stream (Bernoulli) before the solution was released into the wider glass tube.

The breaking of the water stream at the root should have caused additional turbulences as well. At high rates of water flow, the roots tended to tremble depending on their length and mechanical rigidity. Trembling or even shaking was quite intensive with the softer wheat roots (14-d-old and raised hydroponically) compared with the tougher corn roots (S.Q. Zhang, personal communication). Roots were approximately centred in the glass tube. Often, their shape deviated from that of a straight cylinder by slight curvatures. Because of the turbulences in the glass tube, the rates of water flow in the centre of the glass tube should have been close to the measured average value rather than to the higher rates which could be calculated assuming an ideal laminar streaming (Poiseuille's law). The set-up allowed the solution to be quickly pushed out from the glass tube. When changing solutions, the mixing area between old and new solutions should have been no longer than 50 mm. At a root length of 100 mm, this would relate to a time of 0.4 s for a complete exchange of media around a root. This was small compared with the rate of water exchange across roots in osmotic experiments (around 30 s). Hence, the time required for an exchange of media was not rate-limiting during these experiments.

After tightening the silicone seals in steps with the aid of a screw, root pressure (Pr) began to increase and became steady within 1–3 h (Hose et al. 2000). In hydrostatic experiments, pressure relaxations were induced with the aid of the probe to measure the hydrostatic half time (inline image) of radial water exchange across the root (inline image ∝ 1/Lphr; Lphr = hydrostatic root hydraulic conductivity). As described by Hose et al. (2000), hydrostatic relaxation curves were composed of two exponential phases brought about by different rates of changes of Pr with time: the initial rapid phase covered about 80% of the entire pressure (volume) change and was followed by a slow reversible phase (about 20% of the entire change) which is related to concentration polarization effects at the endodermis (Steudle & Frensch 1989). In the present study, the initial phase of hydrostatic relaxation curve was used to measure inline image. In osmotic experiments, the original root medium was rapidly exchanged by a solution containing 320 m m acetone (0.8 MPa of osmotic pressure) in the root medium in addition to the other contents.

From the first phase (water phase) of the biphasic pressure responses, osmotic root hydraulic conductivity (Lpor) was calculated from the osmotic half time (inline image ∝ 1 / Lpor). In order to work out inline image, the portion around the minimum (maximum) was not incorporated (about 10% of the overall pressure change). This portion contains significant interactions between water and solute flows, which tend to increase the rate (Steudle & Tyerman 1983; Tyerman & Steudle 1984). It was not used to work out inline image. From the second phase (solute phase), the permeability of acetone across the root (Psr) was calculated from the half time of acetone flow (inline image ∝ 1/Psr). From the biphasic curves, the reflection coefficient of roots (σsr) was worked out, too. For the calculation of Lpr, Psr and σsr, the reader is referred to earlier publications such as Henzler et al. (1999) and Hose et al. (2000). After re-attaining a stable Pr, the acetone solution was replaced by the original root medium. The root was then treated for 0.5 h with 35 m m H2O2 applied to the root medium. After this treatment, hydrostatic and osmotic experiments were again performed to observe the effects. In the presence of 35 m m H2O2, ABA [(±)-cis-trans-] was added to the medium at a concentration of 1 µm and hydrostatic inline image was observed for 0.5–1 h. Eventually, the medium was exchanged for the original root medium to check whether the effects were reversible. After experiments with a given root, the root was cut at a position close to the seal. When Pr immediately decreased to close to zero and the half times of pressure relaxations became short compared with the original values (less than 1 s versus 4–12 s observed for the intact system in the hydrostatic experiment and 20–60 s in the osmotic experiment), this indicated that root xylem within the seal remained open. Delayed effect or the absence of any effect upon cutting indicated that the root xylem was interrupted during or after tightening. Results from these experiments were discarded (Peterson & Steudle 1993). For a given root, the whole time course of the experiments lasted 8–10 h (including the time to get steady Pr). The Pr (0.1–0.2 MPa) changed within ±0.03 MPa during the experiments.

Cell pressure probe experiments

The same type of cell pressure probe used by Henzler et al. (1999) or Wan et al. (2004) was employed. A root segment (length: 80–120 mm) was fixed by magnetic bars on a metal sledge arranged at an angle of 45°. It was placed on a layer of paper tissue and covered by another layer of paper tissue to keep the root wet. The centre part of the root was lying open at a length of 10 mm, and the solution was running down along the root and the edges of the metal plate. An average rate of solution flow was calculated by dividing the overall water flow of around 800 mm3 s−1 (0.8 mL s−1) by the estimated cross-section of solution flow along the root (most of the flow that was running down the metal sledge) and along the edges of the sledge, which was 4–5 mm2. Hence, the flow rate along the root was at least 0.16–0.20 m s−1. The flow was turbulent and displayed some flickering in shape. Higher rates of solution flow were not possible because higher turbulences at the root surface would have interfered with measurement, with the cell pressure probe tending to cause leakages due to vibrations of the root and/or the tip of the probe's microcapillary.

Using the probe, we punctured cortical cells from the second to fourth layer about 50 mm from the root tip. In the oil-filled microcapillary of the probe, a meniscus formed between cell sap and oil. Cell turgor was rebuilt by gently pushing the meniscus to a position close to the surface of the root. In hydrostatic experiments, pressure relaxations were applied to measure the hydrostatic half time (inline image) of water flow across the cell membrane, which is inversely proportional to the hydraulic conductivity of the cell membrane (inline image ∝ 1/Lph; Lph = hydrostatic cell hydraulic conductivity). To avoid a mechanical inhibition of AQPs, the peak sizes of pressure change were less than 0.1 MPa (Wan et al. 2004). As in the experiments at the root level, osmotic experiments were conducted by adding acetone to the circulating medium at a final concentration of 200 m m. Even at high rates of solution flow along the segments, there may have been reductions in the osmotic concentration of the volatile solute during its passage down the sledge. This was regularly checked with an osmometer (Gonotec, Berlin, Germany). The osmotic half times (inline image ∝ 1/Lpo; Lpo = osmotic cell hydraulic conductivity) were measured from the water phase of responses. Unlike the hydrostatic cell experiments, the half time of water exchange from the cell osmotic experiments referred to a complex barrier that was not as well defined as that in the root pressure probe experiments. The tissue barrier was thinner (cell pressure probe in the second to fourth layer) than that during root experiments. It was somewhat variable depending on which layer was punctured. Hence, only relative changes in Lpo (inline image) rather than absolute values could be worked out, as in the cell experiments using hydrostatic gradients.

Half times of acetone transport (inline image) were measured from the solute phases, which referred to the inverse of solute permeability (Ps) of the same complex barrier between cell and medium rather than just to the membrane of the punctured cell. As for Lpo, relative changes in Ps(inline image) rather than absolute values could be given because of the variability of the thickness of the tissue layer. After treating the root with 35 m m H2O2, the same experiments as in the control were performed in order to observe the effects of H2O2 on water and acetone transport. In some experiments, ABA [(±)-cis-trans-] was added to the medium in the presence of H2O2, resulting in a final concentration of 1 µm of ABA. Following this addition, inline image was observed for 0.5–1 h. To check whether the effects were reversible, the same experiments as in the control were repeated after changing back to the original root medium. Cell turgor was between 0.4 and 0.6 MPa. Within ±0.04 MPa, cell turgor remained constant during the experiments which lasted for 2–4 h for a given cell.


In the presence of 35 m m H2O2, the radial permeability of water across corn roots (Lphr) was substantially inhibited. As for Chara internodes, we attribute this to the existence of reactive oxygen species (ROS), namely *OH, which should have been generated in the presence of Fe2+ in the root (see Introduction). As can be seen from Fig. 1, the hydrostatic half times of root water permeability (inline image) increased from 8.6 to 21 s on average in the presence of 35 m m H2O2 (i.e. Lphr decreased by a factor of 2.5). When removing H2O2 from the medium, the original inline image recovered within 0.5–1 h. In the presence of acetone, which rapidly permeated the root, the osmotic response curves were biphasic as shown in Fig. 1a (Frensch & Steudle 1989; Steudle 1993). This is because roots behave like osmometers that are permeable to both water and solutes (Steudle & Brinckmann 1989). The first phase, during which Pr decreased or increased with a half time of between 20 and 60 s due to an exosmotic/endosmotic volume flow, is dominated by water and is called water phase (Steudle & Jeschke 1983; Steudle 1993). During the solute phase, Pr increased or decreased again because of the passive flow of solute tending to equilibrate the concentration of permeating solutes on both sides of the barrier (medium and xylem), and water follows. Osmotic half times (inline image) measured from the water phase were longer than hydrostatic inline image by a factor of four. This indicated that the Lpor was smaller than the Lphr by the same factor. As inline image, H2O2 treatment increased inline image by a factor of two. The effects were reversible as can be seen in Fig. 1c.

Figure 1.

(a) Typical hydrostatic relaxations of root pressure (Pr) and biphasic osmotic response curves as measured by a root pressure probe in a root excised from corn seedling and treated with 35 m m H2O2. Compared with the control, the hydrostatic half times (inline image) increased by a factor of 2.5. In osmotic experiments with the permeating solute acetone the responses were biphasic. (b) Treatment with 35 m m H2O2 caused anomalous (negative) osmosis (root was swelling in hypertonic solution because the solute entered the root faster than the water could get out). In addition, it increased the half times (inline image) of acetone permeability (Psr) from 190 s to 480 s (on average). (c) Removal of H2O2 from the medium resulted in restoration of root Lpr and Psr and of normal osmosis as in the control.

Because acetone was rather permeable for root cell membranes, the reflection coefficient of roots was close to zero (σsr = 0.04 ± 0.02; n = 5 roots). However, in the presence of H2O2, the osmotic response of the root to the permeating solute reversed its direction. The reflection coefficient of acetone changed from 0.04 ± 0.02 to −0.04 ± 0.01. Although absolute values of reflection coefficients were small, anomalous osmosis was clearly observed; acetone entered the root faster than the water could get out, and Pr was increasing instead of decreasing (see Discussion). This meant that the root was not shrinking but was swelling in the presence of the hypertonic solution (Fig. 1b). Removal of H2O2 from the medium again resulted in normal osmosis as in the control (Fig. 1c). Hence, H2O2 caused a substantial decrease in the water permeability of the root. H2O2 also reduced the permeability of the root for acetone as the half time of acetone permeation increased from 190 to 480 s on average.

In Fig. 2, the relative changes of half times of water flow in response to H2O2 treatment are summarized for the root level. The figure shows that increases in hydrostatic inline image (a) and osmotic inline image (b) were similar. On average, inline image increased by a factor of three; thus, Lphr should have decreased by the same factor. The stress hormone ABA, which has been found to increase transiently root hydraulic conductivity (Hose et al. 2000), had no positive effect on the water permeability of the roots in the presence of H2O2. The osmotic half time inline image increased by a factor of 3.8 when the roots were treated with H2O2. After H2O2 was removed from the medium and roots washed with the control medium, the original half times were restored within 0.5–1 h in both types of experiments.

Figure 2.

Summary of the effects of 35 m m H2O2 and of 1 µm ABA on the half times (a, inline image ∝ 1/Lphr; b, inline image ∝ 1/Lpor) of root hydraulic conductivity (Lpr) (n = 6 roots ± SD). Relative changes are given rather than absolute values to avoid the variability between roots. Treatment of H2O2 increased inline image and inline image by factors of 3 and 3.8, respectively; hydrostatic root hydraulic conductivity (Lphr) or osmotic root hydraulic conductivity (Lpor) decreased by the same factor. The addition of ABA in the presence of H2O2 had no significant effect on inline image. Removal of H2O2 from the medium resulted in a recovery of both inline image and inline image. Different symbols on the top of each column denote significant differences between results (t-test; P < 0.01).

The results from the root level paralleled those found at the level of individual cortical cells. However, at the cell level, the effects on water flow were more pronounced than those at the root level. Treatment with 35 m m H2O2 dramatically inhibited AQP activity and increased half times of water permeability across cell membranes (hydrostatic inline image). In the example given in Fig. 3, hydrostatic inline image increased from 0.8 to 8.0 s (on average), i.e. by one order of magnitude. The effect of removing H2O2 from the medium was completely reversible. As at the root level, the closure of water channels by treatment of 35 m m H2O2 resulted in anomalous osmosis and increased the half times for water and acetone permeation (Fig. 3b). Again, removal of H2O2 from the medium resulted in a reversal to normal osmosis (as in the control; Fig. 3c). As for inline image, a similar effect on osmotic half time (inline image) was found. It is clear that H2O2 treatment was effective in reversibly closing water channels in the cortical cells of corn roots, although function–repair mechanisms are not yet known (see Discussion).

Figure 3.

Typical curves of hydrostatic and osmotic pressure relaxations as measured with a cell pressure probe in an individual cortical cell of corn root as affected by H2O2 treatment. (a) With the permeating solute acetone, responses were biphasic in osmotic experiments. (b) Half times (inline image) of water flow increased by a factor of 10 as compared with the control indicating that water channel activity was substantially inhibited. As at the root level, the cell was swelling in a hypertonic acetone solution (anomalous osmosis) in the presence of 35 m m H2O2. Half times (inline image) of acetone permeability increased from 55 s to 83 s (on average). (c) Removal of H2O2 from the medium resulted in restoration of Lp and Ps to a good approximation and of normal osmosis, as in the control.

Relative changes in the half times of cell Lp found during treatment with H2O2 are summarized in Fig. 4 (hydrostatic experiments). During water channel closure in the presence of H2O2, inline image increased by a factor of nine, indicating that Lph decreased by the same factor. As at the root level, ABA had no positive effect on water permeability across cell membranes in the presence of H2O2. Cell Lph was recovered within 15–30 min after the removal of H2O2 from the medium. As the roots recovered, results indicated that H2O2 treatment did not simply cause an overall damage on membranes (Aroca et al. 2005). Furthermore, in case of an unspecific damage, there should have been an increase, rather than a decrease of Lp and Ps upon treatment.

Figure 4.

Summary of the effects of 35 m m H2O2 and of 1 µm ABA on the half times (inline image ∝ 1/Lph) of cell hydraulic conductivity (n = 6 cells ± SD). As in Fig. 2, relative changes are given to avoid the variability between cells. The half time (inline image) of cell hydraulic conductivity (Lph) increased by a factor of nine as compared with the control, indicating that the activity of aquaporins (AQPs) was substantially inhibited by treatment with H2O2. In the presence of H2O2, addition of ABA had no effect on inline image. Removal of H2O2 from the medium resulted in a recovery of inline image. Different letters on the top of each column denote significant differences between results (t-test; P < 0.01).

Hydrogen peroxide increased the half times of acetone permeability at the level of both cortical cells (Fig. 5a) and entire roots (Fig. 5b) by factors of 1.5 and 2, respectively. This meant that the permeabilities of acetone were reduced by the same factors. When H2O2 was removed from the medium, the original permeabilities were recovered. These results are similar to previous findings that AQPs in Chara are not ideally selective for water but allow, to some extent, the passage of small organic solutes (Hertel & Steudle 1997; Henzler et al. 2004).

Figure 5.

Summary of the effects of 35 m m H2O2 on half times (inline image ∝ 1/Psr; inline image ∝ 1/Ps) of acetone permeability across roots (a) and ‘cells’ (b) (n = 6 roots or cells ± SD). Relative changes are given to avoid the variability between roots or cells. Treatment with H2O2 doubled inline image (Psr decreased by a factor of two) and increased inline image of acetone permeability by 50% as compared with the control, indicating that a substantial amount of acetone was permeating across aquaporins (AQPs) which were not completely selective for water. Both inline image and inline image were recovered after the removal of H2O2 from the medium. Different symbols on the top of each column denote significant differences between results (t-test; P < 0.05).


The results indicate an oxidative gating of AQPs in the membranes of the root cells of young corn seedlings which, in turn, caused a reversible reduction of the overall water permeability of roots. At least qualitatively, the effects are similar to those obtained recently for the plasma membrane of Chara corallina (Henzler et al. 2004). As for the isolated internodes of Chara, channel closure in root cell membranes caused anomalous osmosis in the presence of a rapidly permeating solute (acetone), that is, closure of AQPs caused the solute to move faster than the water and the reflection coefficient became negative (σsr = −0.04 versus σs = −0.50 for Chara; Henzler et al. 2004). The responses at the cell level were more pronounced than those at the root level. Overall, the results suggest an oxidative gating of AQPs in corn roots similar to that already found for Chara.

It may be argued that during the experiments at the cell and root level, external unstirred layers played an important role or may have been even dominant during water and solute movement. However, the high rates of solution flow around roots in both types of experiments (cell and root pressure probes) guaranteed that the effects of external unstirred layers should have been small. Water around roots and root sections was in turbulent motion (see Materials and methods). Hence, the thickness of unstirred layers should have been no more than 50 µm (Steudle & Tyerman 1983). According to the Einstein–Smoluchowski relation for a plane sheet, this would refer to a half time of 0.6 s for the complete equilibration with the external solute acetone [D = 1.1 × 10−9 m2·s−1; t1/2 = 0.281 × (5 × 10−5)2 / (1.1 × 10−9); Jost 1960]. The rates of water exchange across the unstirred layer should have been much faster than those across the root in the osmotic experiments (inline image = 20–60 s in the untreated root). A thickness of 50 µm would refer to an acetone permeability of 2.2 × 10−5 m·s−1, which is much larger than that measured for cell membranes (Steudle & Tyerman 1983; Henzler et al. 2004). Hence, the unstirred layer still adhering to the root even in the presence of vigorous stirring should not contribute much during the root pressure probe experiments. In the hydrostatic experiments with individual cortical cells, unstirred layer effects should have been even smaller. In these experiments, only a small amount of water is moved across the membrane forming a layer of only a fraction of a micron (sweep-away effect) (Dainty 1963; Steudle, Smith & Lüttge 1980; Steudle & Tyerman 1983; Steudle 1993).

In Chara internodes, treatment with *OH (as produced by the Fenton reaction in the presence of a fraction of a millimole of H2O2) reversibly reduced cell Lp by more than 90% (Henzler et al. 2004). It has been verified that *OH rather than H2O2 was the inhibiting agent. In the absence of Fe2+ in the medium, Chara cells tolerated H2O2 at concentrations of up to 350 m m without affecting cell Lp and cell turgor (Henzler & Steudle 2000).

In the present study, the equivalent test could not be made because the roots should have contained sufficient Fe2+ or cations from other transition metals (see Introduction; Chen & Schopfer 1999). In the presence of 18 µm of FeNaEDTA (much less than that in the experiments with Chara), a concentration of H2O2 as large as 35 m m was required to cause effects similar to those in Chara. This may point to the fact that *OH radicals produced in roots were also used in reactions other than the inhibition of AQPs, or that AQPs of roots were less sensitive to *OH radicals than those of Chara internodes, or both. For Chara and root cells, changes in Lph were reversible. For Chara, it has been proposed that, because of the high reactivity of radicals, *OH radicals were most likely produced close to the cell membranes (Henzler et al. 2004). This may also be true for corn roots. Besides the idea that AQPs were directly attacked by *OH radicals, there may be another mechanism in which C=C double bonds of the plasma membrane were attacked by *OH, resulting in the formation of aggressive radicals that attacked AQPs laterally from inside the bilayer. Both types of chemical attacks may have resulted in conformational changes of the water channel proteins and their reversible closure. Which type of mechanism is valid cannot be decided from the present results. However, there may have been a similar mechanism by which chemical alterations were removed, most likely by the reduction of oxidized groups of AQPs (Henzler et al. 2004). Alternatively, H2O2 is thought to be a messenger molecule involved in signal transductions during which plants suffer biotic or abiotic stresses (Pastori & Foyer 2002; Xiong et al. 2002). Hence, the regulation of water channel activity by oxidative signalling in the presence of ROS could play a role as part of a downstream reaction to stresses, such as low temperature, drought or high light intensity or during a pathogen attack (Wojtaszek 1997; Pei et al. 2000; Neill, Desikan & Hancock 2002), probably by eliciting the activation of Ca2+ permeable channels in cell membranes, which in turn, block water channels through intracellular Ca2+ signalling (Gerbeau et al. 2002; Mori & Schroeder 2004).

The addition of a rapidly permeating solute to the root medium resulted in biphasic osmotic response curves (Steudle 1993). This type of response has been demonstrated in the past for different roots and quite a number of solutes including those that rapidly permeate membranes (Steudle, Oren & Schulze 1987; Steudle & Brinckmann 1989). In the biphasic responses, there is a water phase due to an exosmotic volume flow (largely water) which causes a pressure decrease. This is followed by a solute phase due to the passive flow of the solute and water is again taken up (pressure increase). Responses of roots are similar to those of isolated cells, although not identical (Steudle et al. 1987). At the cell level, there are cases in which turgor pressure increases in the presence of a hypertonic solution, an occurrence known as anomalous osmosis (Steudle & Henzler 1995; Henzler et al. 2004). To date, anomalous osmosis could only be induced in the presence of rather ‘exotic’ solutes that exhibit a permeability similar to that of water (such as acetone). The phenomenon refers to the striking situation that cells do not shrink but swell in hypertonic media. Here, we present evidence that anomalous osmosis may also be demonstrated in roots, that is, for an entire organ and in the presence of a somewhat complicated osmotic barrier. At first glance, anomalous osmosis takes place when solutes enter the cell or root faster than the water can get out, as when AQPs are closed and osmolytes are more rapidly permeating the membrane or barrier. However, this simple interpretation in terms of a change in the ranking between water and solute permeability cannot completely explain the story. The biphasic pressure/time curves (or volume/time curves) in the presence of a permeating solute at the cell or root level (Steudle & Tyerman 1983; Steudle et al. 1987) can be described by:


where V(t) = cell volume; Vo = cell volume at t = 0; P(t) = cell turgor pressure; Po = cell turgor pressure at t = 0; A = cell surface area; inline image = change in the external osmotic pressure; kw = Lp·A(ɛ + inline image) / V= rate constant of water flow; and ks = Ps·A/V = rate constant of solute flow. Because of the presence of AQPs, water movement across membranes or roots is usually much faster than that of solutes, that is, kw > ks. If solute transport is faster than that of water (i.e. ks > kw), this will change the sign of the term within the brackets in Eqn 1. However, ks > kw will also change the sign of the denominator on the right side of Eqn 1 and the effect of a change from kw > ks to ks > kw will cancel. Hence, the anomalous osmosis is not due to the relative of kw versus ks. According to Eqn 1, it is due to a change in the σs from positive to negative values. Although often correlated with the permeabilities of water and solutes (Ps and Lp), the σs is an independent parameter required to describe osmosis besides Lp and Ps.

In a homogenous membrane that lacks pores (which provide the only way for a direct interaction between water and solute flow), the σs should be related to the water and solute permeability (Dainty 1963; Steudle & Henzler 1995) by:


where inline image is the partial molar volume of the solute; Pd is the diffusional water permeability; and inline image is the molar volume of liquid water. The second term on the right side of the equation represents the contribution of solute flow to the overall volume flow. A comparison of measured and calculated values of reflection coefficients shows that measured values are usually substantially smaller than those calculated from Ps and Lph (e.g. Table 1 of Steudle & Tyerman 1983). In the past, this has been attributed to the existence of water-filled pores and to a frictional interaction between water and solute as they cross the membrane (Dainty 1963; Steudle & Tyerman 1983):


In Eqn 3, the last term on the right side denotes the frictional interaction; inline image is the partition coefficient of solute ‘s’ between membrane pores and the membrane; inline image represents the frictional interaction between solute and water in the pores; and inline image is the interaction between solutes and the wall of the pores. The frictional term, inline image, describes the per mole force, which acts on water and solute molecules when they pass the pore in opposite directions during an osmotic experiment in the presence of a permeating solute. The model assumes rather wide pores where water and solutes may pass each other. However, recent evidence showed that most of the water uses narrow and rather selective AQP pores where the water is aligned in single files (no-pass pores, e.g. Jung et al. 1994; Maurel 1997; Ren et al. 2001). It turned out that AQPs allowed the passage of small uncharged solutes such as acetone and a limited amount of other solutes (Steudle & Henzler 1995; Henzler et al. 2004). The coupling in single-file pores tends to reduce reflection coefficients, but the effects should be small according to the limited stoichiometric coupling between solutes and water during their passage across the membrane (Finkelstein 1987; Hertel & Steudle 1997). How then do we obtain negative σs and anomalous osmosis? According to Eqns 2 and 3, the membrane is looked at as a homogenous structure, i.e. water and solutes are just passing through the pores, and the passage across the parallel bilayer (or other structures) is not taken into account explicitly. This model cannot explain negative σs for lipohilic solutes upon channel closure (inline image = 0). However, the finding is explained when the composite structure of membranes (or the composite transport model of membranes) is taken into account (Steudle & Henzler 1995). Based on this model, channel closure could result in a negative σs according to Eqn 2, provided that inline image·Ps > inline image·Pd holds for a lipophilic solute such as acetone. When the porous passage (AQPs) is closed, the overall transport properties of the membrane should assume those of the bilayer (or of the rest of the membrane), and the overall σs could be negative for these solutes, as found (Henzler & Steudle 1995; Henzler et al. 2004). Hence, the water/solute movement across the membrane is explained in terms of the composite nature of membrane transport following the basic treatment of Kedem & Katchalsky (1963a) for parallel transport elements (KK concept).

In tissues such as corn roots, the KK concept should apply as well (Steudle et al. 1987; Steudle & Peterson 1998; Tyerman et al. 1999). Unlike in the membrane, series arrays would have to be considered in roots besides the parallel such as different tissues (rhizodermis, cortex, stele) or special barriers such as the exodermis and the endodermis with its Casparian bands. Again, basic concepts should apply both for the parallel and the series arrangements (Kedem & Katchalsky 1963a,b; House 1974). With respect to the parallel elements, the apoplast has to be considered besides the other two pathways. One would expect a σs of close to zero for acetone in the apoplast and a relatively high diffusional permeability along this structure. In untreated cells of the root cortex, the σs should be small but positive (for Chara, the σs of acetone was 0.15; Ye et al. 2004). According to the parallel arrangement of pathways, the basic theory predicts that the overall σs should have been between 0 (apoplast) and 0.15 (cell-to-cell passage), as found. The σs of different pathways should have contributed according to their individual conductances, which, however, are not known. The closing of channels in root cells should have caused a negative σs (in Chara, this resulted in a substantially negative reflection coefficient for acetone, σs = −0.50). In the root, one would expect an overall value of between zero and this value, as also found. Hence, the measured switching between normal and anomalous osmosis is explained by the composite transport model, as are the rather low absolute values of σsr. The same refers to the osmotic experiments with cortical cells, which differed from those of entire roots just by the somewhat variable thickness of the layer of tissue around them.

According to the composite transport model for the radial transport of water across the root cylinder (Steudle 2000, 2001), root Lpr very much depends on how intensively the parallel apoplastic and cell-to-cell pathways are used by the water. The cell-to-cell component would be determined to a large extent by AQP activity. It has been shown that the relative contribution of the two components is highly variable depending on the nature of the driving force (osmotic versus hydrostatic) and root anatomy (e.g. Steudle & Peterson 1998; Lee et al. 2005a). In the presence of hydrostatic pressure gradients, water flow should be largely around protoplasts (i.e. apoplastic) because this path represents a low hydraulic resistance (high Lphr). On the other hand, water flow in the presence of osmotic gradients is low in that osmotic driving forces cause water movement largely across membranes. An osmotic water flow across the root has to pass many membranes, which then results in Lpor smaller than Lphr (Zhu & Steudle 1991; Steudle & Frensch 1996). Under conditions of salinity stress, which inhibits AQP activity of corn root cells, the effects on cell Lph were substantially bigger than those on root Lphr (Azaizeh et al. 1992). However, the results of Azaizeh et al. (1992) showed that inhibition of cell Lph also affected hydrostatic root Lphr, indicating a contribution of membranes along the path (Steudle 1992). On longer terms, the apoplastic path may be affected by anatomical changes such as the formation of barriers (Zimmermann et al. 2000; Lee et al. 2005a). In the short-term treatment with H2O2, it is most likely that the transport step in the endodermis included a bigger membrane component than in the rest of the root (Zimmermann & Steudle 1998). Nevertheless, the effects in response to the oxidative gating were much bigger at the cell (reduction by a factor of nine) than at the root or tissue level (reduction by a factor of three).

The finding that the fold change of osmotic Lpor caused by the treatment was similar to the fold change in hydrostatic Lphr (3.8-fold versus 3-fold) seems to contradict the interpretation according to composite structures, which is usually in terms of parallel transport elements. During the hydrostatic experiment, water flow should largely bypass protoplasts along the apoplast, which should not be affected by the H2O2 treatment. During the osmotic experiment, however, it should be largely along the cell-to-cell path. If the latter is true, we may calculate Lpor, assuming that all the water moved across cell membranes arranged in concentric cylinders and not in the apoplast. For each cell layer, two membranes would have to be crossed. The overall Lpor would relate to individual cell Lp (Steudle & Brinckmann 1989) by:


where n is the number of membrane layers which would have to be crossed by the water (twice as many as cell layers); ro is the radius of the root; and ri is the radius of the ith membrane layer. For a typical corn root of 1 mm in diameter (such as those used in the present study), 12–14 cell layers would have to be crossed by the water. Taking the  average  diameters  of  cells  in  different  layers  from  an

earlier  paper  (Table 2  of Zhu  &  Steudle  1991), inline image  was

calculated to be 74. Using a typical cell Lph of 6.0 × 10−6 m· s−1·MPa−1 (equivalent to inline image = 0.8 s), this should have resulted in Lpor of 8.0 × 10−8 m·s−1·MPa−1. This value is close to that measured with the root pressure probe (Lpor = 7.0 × 10−8 m·s−1·MPa−1; equivalent to a inline image = 35 s). The estimate supports the view of a largely cell-to-cell movement of water in osmotic experiments.

On the other hand, the passage of water during a hydrostatic root experiment is not completely free of membrane components, namely, in the endodermis or the exodermis where Casparian bands tend to interrupt the apoplastic flow (Steudle & Peterson 1998; Zimmermann & Steudle 1998; Zimmermann et al. 2000). The present findings agree with this view. When endodermal cell Lp can be identified with that of other cortical cells, we would expect a reduction of Lpor by a factor of nine (Fig. 3) provided that the endodermis represents the dominating hydraulic resistance and the Casparian bands completely interrupt water flow (according to the classical view of the function of the bands). However, the latter is not true (Ranathunge, Steudle & Lafitte 2005), and the finding of a reduction of only 3.8 instead of 9-fold is understandable. The similar fold change in hydrostatic and osmotic root Lpr does not mean that, in both types of experiments, water flow was predominantly from cell to cell and that the hydrostatic root Lpr of untreated roots was larger by a factor of four because of unstirred layers in the apoplast. If this were true, then the apoplastic unstirred layers in the root cortex and stele would have been bypassed by the high cell-to-cell permeability. A rigorous treatment of apoplastic diffusional unstirred layers within roots has shown that they would contribute to less than expected to the overall root Lpr (Steudle & Frensch 1989). These authors assumed thicknesses of unstirred layers of as large as the entire cortex (300 µm) and a stelar unstirred layer of as large as 50 µm. Assuming a reduction of solute diffusibility within the root apoplast by a factor of 5–20, they arrived at the conclusion that the contribution of unstirred layers was as small as 7%. The apoplastic diffusive mobility should be similar for the solutes used by Steudle & Frensch (1989) as for the acetone used here. It has to be kept in mind that, during composite transport, the parallel pathways interact with each other by a rapid equilibration of water called ‘local equilibrium’ (Steudle 1992). Hence, treating the apoplast on both sides of the endodermis as an ordinary diffusional unstirred layer would be premature.

In plant roots, the stress hormone ABA is produced under unfavourable conditions such as water shortage (Zhang, Schurr & Davies 1987). It is transported to the shoot as a signal that induces closure of stomata in leaves. It has been reported that in corn roots, ABA increased the root and cell hydraulic conductivity (Freundl et al. 1998, 2000; Hose et al. 2000, 2001). Recently, it was found that ABA has a positive effect on AQP activity when channels were closed by mechanical stimuli or during low temperature (Wan et al. 2004; Lee et al. 2005b). It was concluded that ABA tended to restore the original open state of channel proteins by a mechanism not yet known. However, in the present study, ABA had no such effect when AQPs were blocked by oxidative gating. This may be due to the fact that the gating mechanism of AQPs by oxidation is different from that by mechanical stimuli or low temperature. We hypothesize that ABA could not recover chemical modifications of AQPs during the treatment with H2O2 or *OH radicals. This may require a biochemical process (reduction of oxidized AQPs) rather than just a physical action (change of activation energy of the transition between different conformational states to reopen closed channels).

In conclusion, the closure of AQPs in young corn roots caused responses in cell Lp and root Lpr that were similar to those found in Chara internodes. Water permeabilities at the root and cell level were substantially and reversibly reduced, indicating an inhibition of AQP activity by oxidative stress (*OH radicals and/or H2O2). As in previous experiments with roots, the inhibition of AQPs at the cell level resulted in a decrease of cell Lph which was larger than that of the root or tissue Lphr. Unlike in Chara, a much higher H2O2 concentration was required in corn roots. This may be due to the fact that AQP responsiveness in roots was smaller or that *OH radicals were also used in other processes, or both. As for the Chara system, anomalous osmosis (negative σs) could be reversibly induced during oxidative stress in the presence of the rapidly permeating solute acetone which should have largely used the bilayer to pass through the root cylinder. Oxidative stress inhibited cell Lph by a factor of nine, which was larger than the inhibition at the root level (factor of three). Differences in the inhibition are readily explained in terms of the root's composite transport model, which also explains the switching between normal and anomalous osmosis in the presence of the rapidly permeating solute. As in the application of other stresses, the hydrostatic root Lphr was also affected by oxidative treatment. This indicated that the apoplastic passage was partially interrupted, most likely at the endodermis. Unlike the gating by mechanical stimuli (Wan et al. 2004) or by low temperature (Lee et al. 2005a,b), the stress hormone ABA had no ameliorative effect on restoring the original conformation of channel proteins and reopening closed channels. This is possibly because the oxidative gating in the presence of *OH radicals caused a chemical modification of AQPs. The precise mechanism of the oxidative gating of AQPs in the presence of the ROS (*OH or H2O2) is not yet understood. Three alternatives may be possible: (1) AQPs were directly attacked (oxidized) by *OH; (2) AQPs were oxidized by other aggressive radicals which were subsequently produced by *OH; and (3) H2O2 may have elicited changes in cytoplasmic Ca2+ concentration through cell signalling cascades resulting in channel closure. In all three cases, AQP activity could be regulated by an oxidative gating or signalling initiated by the presence of H2O2 or *OH radicals. There may be a common interaction between the redox state (oxidative stress) and water relations (water stress) in the life of plants.


The authors would like to thank Burkhard Stumpf (Department of Plant Ecology, University of Bayreuth) for his expert technical assistance.