Photosynthesis and biomass production of plants are controlled by the water status of the soil. Upon soil drying, plants can reduce water consumption by minimizing transpiration through stomata, the closable pores of the leaf. The phytohormone abscisic acid (ABA) mediates stomatal closure, and is the assigned signal for communicating water deficit from the root to the shoot. However, our study does not support ABA as the proposed long-distance signal. The shoot response to limited soil water supply is not affected by the capacity to generate ABA in the root; however, the response does require ABA biosynthesis and signalling in the shoot. Soil water stress elicits a hydraulic response in the shoot, which precedes ABA signalling and stomatal closure. Attenuation of the hydraulic response in various plants prevented long-distance signalling of water stress, consistent with root-to-shoot communication by a hydraulic signal.
Biomass production by plants is regulated by water availability (Schulze, 1986). Transpiration and the gas exchange of CO2 and oxygen are controlled by adjusting the stomatal aperture of pores formed by guard cells in the leaf epidermis (Hetherington and Woodward, 2003). Plants can adjust to widely varying water potentials (ψ), ranging from almost zero in well-watered soils down to −8 MPa in arid areas (Odening et al., 1974). A limited water supply in the soil is sensed by roots and communicated to the shoot to allow timely adjustment of the plant’s transpiration (Schulze and Hall, 1982). According to current understanding, abscisic acid (ABA) is the long-distance signal that communicates water stress from the root to the shoot (Wilkinson and Davies, 2002).
Abscisic acid plays a pivotal role in adjustment of plants to abiotic stress conditions (Christmann et al., 2006). It mediates stomatal closure, but also regulates plant development in the absence of stress (Cheng et al., 2002). The sesquiterpene ABA is derived from the C15 compound xanthoxin that is generated by cleavage of a carotenoid (Schwartz et al., 2003), and the reaction is stimulated by drought conditions (Qin and Zeevaart, 1999). The enzymes that convert xanthoxin into ABA are expressed in the vascular parenchyma cells of roots and shoots (Cheng et al., 2002; Koiwai et al., 2004). Evidence for root-derived ABA as a long-distance signal has been obtained from split-root experiments with whole plants in which only one part of the root system experienced water deficit (Wilkinson and Davies, 2002). Likewise, maintaining the water status in leaves by pressurization of drought-exposed root systems did not prevent stomatal closure in herbaceous species (Wilkinson and Davies, 2002). However, grafting of tomato shoots onto root stocks with deficiency in ABA biosynthesis indicated that root-generated ABA is not required, or, alternatively, that transported ABA from the shoot may compensate for ABA deficiency in roots (Holbrook et al., 2002). ABA can be stored as physiologically inactive ABA–glucose conjugates from which ABA is released by a specific hydrolase that is activated under water stress (Lee et al., 2006). However, the levels of ABA glucose ester are too small for ABA release from this conjugate to contribute significantly to the overall increase in ABA during water stress (Neill et al., 1983; Priest et al., 2006). Distinct ABA perception and signal transduction control stomatal aperture, seed germination, growth and flowering time (Finkelstein et al., 2002; Mishra et al., 2006; Razem et al., 2006; Schroeder et al., 2001; Shen et al., 2006).
The function of ABA as a long-distance signal that communicates water shortage from the root to the shoot is still not clear. In this study, we used Arabidopsis and a non-invasive imaging system for ABA action (Christmann et al., 2005) to examine the model. Our data do not support ABA as the long-distance signal controlling water-deficit responses in the shoot; rather we present evidence for a hydraulic signal in root-to-shoot communication of water stress.
ABA pools and ABA signalling in response to root water stress
We studied the dynamics of the action of ABA during water stress throughout the plant by non-invasively visualizing ABA signalling via luciferase expression in Arabidopsis. Seedlings water-stressed by exposure of roots to a medium of high osmotic concentration showed prominent ABA signalling in shoot tissue, namely guard cells (Christmann et al., 2005). Under such conditions of low water potential (−1.0 MPa), a weak activation of ABA signalling was observed in root cells. Responses were localized to the proximity of the vascular bundles (Figure 1a), and were absent in control seedlings exposed to high water potential (−0.2 MPa) as well as in the Arabidopsis mutants aba2-1 (Figure 1a) and abi1-1 (not shown) that are deficient in ABA biosynthesis and defective in ABA signalling, respectively (Gonzalez-Guzman et al., 2002; Koornneef et al., 1984; Léon-Kloosterziel et al., 1996; Meyer et al., 1994).
In addition to the plant’s capacity to synthesize and transduce the ABA signal, steady-state stomatal apertures depended on the perirhizal water potential. Complete stomatal closure required a water potential (ψ) <−0.8 MPa in the substratum. Adjustment of the water potential using either mannitol (Figure 1b) or polyethylene glycol (not shown) yielded identical responses, with detectable closing of stomata 1.5 h after onset (Figure 1c). Feeding 30 μM ABA to roots in the absence of water stress activated ABA signalling (Figure 1a) and induced stomatal closure with kinetics comparable to those of water-stressed plants (Figure 1c). ABA feeding also rescued stomatal closure in aba2-1 at low water potential of the substratum (Ψ = −1.0 MPa, data not shown).
Water deficit at the roots resulted in an approximately tenfold increase in ABA levels (Figure 2a) and a comparable ABA-responsive reporter gene expression in the shoot (Figure 2b). As seen by in vivo imaging, ABA signalling was confined to the stele of water-stressed roots. However, the levels were too low to detect significant changes in cell-free extracts (Figure 2a,b). It is possible that ABA may have leaked into the substratum. During 24 h of stress treatment, each seedling delivered on average 30 fmol of ABA to the medium (Table 1). Deuterated ABA was added to the substratum as an internal standard to correct for ABA recovery. During the stress exposure, shoot ABA levels increased by approximately 290 fmol, indicating that phytohormone leakage is significant but accounts for less than 10% of the total ABA in water-stressed seedlings (Table 1). Thus, our failure to observe increased ABA pools in the root may be attributable to some ABA leakage under the experimental conditions. The much higher ABA content in the shoot, however, made us question whether root-to-shoot signalling of water stress does indeed rely on a supply of ABA from the root.
Table 1. Abscisic acid (ABA) levels in substratum, root and shoot in response to low water potential (Ψ = −1.0 MPa/24 h) in the substratum. Seedlings of Arabidopsis were stressed by exposure of roots to mannitol-containing medium (Ψ = −1.0 MPa) for 24 h (water stress) or to non-stress medium (control) prior ABA analysis in extracts of medium, roots and shoots
ABA level (fmol)
Induction (fold increase)
aCalculated per seedling.
30.3 ± 4.3
15 ± 4.1
14.1 ± 9.0
5.4 ± 3.5
295.2 ± 17.7
Shoot-derived ABA is necessary and sufficient for stomatal closure
In order to test the hypothesis that ABA acts as a long-distance signal, we generated reciprocal grafts between wild-type and aba2-1 (ABA-deficient) mutant reporter lines. Grafted plants containing a wild-type-derived scion on root stock of wild-type or ABA-deficient plants closed their stomata in response to perirhizal water shortage (P < 0.001), resulting in mean stomatal width-to-length aperture ratios of 0.02 (Figure 2c). Irrespective of the root’s capacity to generate ABA, stomata of grafts containing an ABA-deficient scion were unable to respond to water deficit (P < 0.001). Stomata remained open (mean aperture ratio of 0.59), and no ABA-dependent reporter induction was detectable (Figure 2d). From these results, we concluded that (i) root-inflicted water stress elicits shoot ABA responses in the absence of root-generated ABA, and (ii) shoot-derived ABA is necessary and sufficient for stomatal closure. The data do not support the model of root-derived ABA translocated from water-stressed roots to the shoot as the long-distance signal.
Attenuation of root-to-shoot water stress signalling
We then explored the nature of the long-distance signal, which could be electrical, chemical or hydraulic in nature. According to the cohesion/tension theory for the ascent of sap in plants, the capillary water of the xylem is part of a continuous hydraulic system that links absorbing surfaces of the roots to evaporating surfaces in the leaves, and hydraulic forces are fairly rapidly transmitted from roots to shoots (Steudle, 2001; Wei et al., 1999). Hence, the hydraulic system has the potential to provide an effective long-distance relay of the plant’s water status (Malone, 1993). To test the potential role of hydraulic signals, we attempted to attenuate signalling of water stress by providing water directly to leaves of water-stressed plants. There was no closure of stomata in response to root-experienced water stress when water was placed directly on leaf surfaces of Arabidopsis seedlings, and ABA-enhanced reporter expression was attenuated (Figure 3a,b). In the root, a marginal activation of ABA-dependent reporter expression was observed upon attenuation of water stress in the shoot. The water feeding performed may intercept the relay of chemical or electric signals or compensate the hydraulic signal. Exogenous application of 30 μm ABA fully rescued stomatal response of water-fed leaves. Six hours after the onset of water stress in roots, the stomata were closed and aperture ratios were indistinguishable from those of seedlings that were subjected to water stress and not water-fed (aperture 0.02 ± 0.07, n = 45 stomata). The data show that root-to-shoot transport of a chemical signal was not blocked and that leaf stomata responded to transported ABA. In order to rule out the possibility that water feeding interfered with root-to-shoot signalling by a water stress-induced increase in the pH of the apoplast (Wilkinson and Davies, 2002), microdroplets (0.4 μl) of pure water were repeatedly administered to the air-exposed leaf surface of root-stressed plantlets to avoid dilution of a pH signal. In addition, microdroplets of solutions adjusted to various pH values (pH = 4.8 to 8.0 with increments of 0.4 units), including those observed in the xylem sap from droughted plants (pH around 7.0), were applied. Such droplets were sufficient to attenuate the ABA responses in leaves irrespective of their pH. Consistent with a hydraulic signal, the stomatal response depended on the water potential of the shoot-fed solution (Figure 3c). Water potentials of the feeding solution close to or more negative than the leaf water potential (–0.6 ± 0.05 MPa) did not attenuate the response, while more positive values partially or fully suppressed it (Figure 3c). Hence, conditions that are predicted to allow net water influx into leaf tissue, which presumably proceeds via the hydathodes, reduced or abolished the closing of stomata. The data therefore favour a hydraulic over an electric or chemical signal.
Detection of a hydraulic signal in leaves in response to root water stress
Transpiration under conditions of water shortage can result in high tensions of almost 10 MPa in the xylem, affecting the turgor pressure of living cells in the vicinity of the xylem (Steudle, 2001), and extensive water deficits lower cell turgor throughout the plant (Hsiao, 1973). Mesophyll cells of well-watered plants revealed a positive cell pressure (turgor) of approximately 0.38 ± 0.04 MPa. Severe water stress reduced the turgor pressure of the leaf cell to zero within 200 sec (Figure 4a). Water deficit was imposed on whole plants by lowering the water potential of the soil using a sorbitol solution (−0.8 MPa). Drainage of sorbitol from the soil by excessive watering reversed the effect and almost recovered the initial turgor values (0.31 ± 0.10 MPa). Similar changes were observed in the aba2 and abi1 mutants (data not shown), and therefore are independent of ABA biosynthesis and signalling. Compared to changes in turgor, stomatal responses were delayed, and full closure resulted after approximately 1 h (Figure 4b).
When leaves were fed with water, a decrease in soil water potential sufficient to elicit a strong response as described above, now only generated a minor change of turgor (Figure 4c). Under these conditions, mesophyll cells remained turgid (turgor approximately 0.3 MPa), and stomata did not close. Removal of sorbitol from the root system recovered the initial turgor of mesophyll cells. However, when the osmotic potential of the sorbitol solution was lowered to −2 MPa and leaves were fed with water, there was a complete loss of turgor, and a subsequent closure of stomata occurred. Attenuation of the hydraulic signal in mesophyll cells shifted the soil water potential required for half-maximal turgor loss from –0.28 ± 0.02 MPa to potentials that were three times more negative (ψ = −0.93 ± 0.12 MPa; Figure 4d).
In order to extend our findings to other plants, seedlings of maple (Acer pseudoplatanus) and beech (Fagus sylvatica) were analyzed for the importance of hydraulic signals in the stomatal response to water stress. Exposing roots to water shortage resulted in stomatal closure in the shoot, unless the shoots were provided with water via the cotelydons or the leaves (Table 2). However, replacing water with solutions of low water potential in the feeding experiment did not prevent stomatal closure. Taken together, the data show that water stress applied to the root generates a hydraulic signature in the shoot. The response was independent of ABA and preceded stomatal closure. The data imply that a hydraulic signal functions in the long-distance communication of water status.
Table 2. Attenuation of stomatal responses to perirhizal water deficit, measured in terms of the stomatal aperture (width vs. length)
Ambient air (AIR)/Murashige and Skoog (MS)
AIR/osmotic stress medium (OSM)
Seedlings of Arabidopsis, beech and maple were in contact at the root with solidified MS medium or osmotic stress medium (Ψ = −1.0 MPa, OSM). The shoot was either exposed to air (AIR/MS or AIR/OSM) or was in contact with solidified medium (MS/OSM or OSM/OSM). Values are for stomatal apertures (n = 90) after one day of exposure.
0.54 ± 0.14
0.01 ± 0.02
0.53 ± 0.10
0.01 ± 0.05
0.51 ± 0.16
0.01 ± 0.06
0.59 ± 0.14
0.03 ± 0.07
0.49 ± 0.15
0.03 ± 0.07
0.49 ± 0.15
0.02 ± 0.06
Water stress in the presence of ongoing photosynthesis puts a major strain on plants. The dilemma is whether to keep stomata closed and combat the lack of the electron acceptor CO2, or to open stomata and tolerate water loss by transpiration. Hence, control of stomatal aperture is based on a regulatory network in which several input signals for opening and closing such as CO2 deficiency and ABA are integrated (Hetherington, 2001; Schroeder et al., 2001; Young et al., 2006). Optimal regulation of leaf transpiration requires the integration of the soil water status. Plants are able to respond rapidly to transient water shortage at midday by closing the stomata and causing midday suppression of photosynthesis (Schulze and Hall, 1982). Such a short-term adjustment is difficult to reconcile with the long-distance transport of a chemical signal in trees.
Transport of a chemical signal from roots to the canopy of tall trees may take more than a day based on the transpirational streaming in conifer xylem with transport velocities of up to 2 m/h (Zimmermann and Brown, 1971). During night or water shortage, transpiration is reduced, and, simultaneously, basi-acropetal transport in the xylem is also reduced (Jackson, 1997). The grafting analyses with ABA-deficient Arabidopsis clearly showed a requirement for ABA in the response to soil-borne water stress, but not as the long-distance signal. ABA biosynthesis in the shoot was necessary and sufficient to mediate stomatal closure of plants water-stressed at the roots, confirming similar grafting experiments with tomato (Holbrook et al., 2002). The detailed analysis of ABA action during the stress response disclosed some ABA signalling in the root stele. A measurable increase in root-localized ABA was not found, and ABA leakage into the substratum may have contributed to this finding. In the shoot, a 10-fold up to more than 20-fold induction of ABA levels was observed during water stress, paralleled by a strong activation of ABA signalling. Neither the constraints on transport velocity in the transpiration stream or our experimental data favour ABA as the root-to-shoot signal. The identification of a change in turgor pressure of mesophyll cells within minutes of root-evoked water stress provided evidence for the involvement of a hydraulic signal.
Stomatal closure in response to water stress required both ABA and a hydraulic signal in the shoot. Attenuation of the hydraulic response prevented ABA signalling and subsequent physiological responses. The attenuation was overcome by feeding ABA to the root system, compatible with a single response pathway of ABA and the hydraulic signal. Generation of the hydraulic signal neither depended on ABA biosynthesis nor on ABA signalling. In response to water stress, the mutants aba2 and abi1 revealed a hydraulic response comparable to that of wild-type; however, the stomata did not close in the mutants. Taken together, the data reveal a signal pathway in which ABA acts downstream of the hydraulic signal in communicating water stress between root and shoot.
In earlier studies on root-to-shoot communication of water shortage, stress was imposed gradually and no changes in leaf water relations were detected (Wilkinson and Davies, 2002). In our study, the stress was evoked suddenly and enabled us to monitor rapid root-to-shoot transmission of a hydraulic signal in the intact plant. The experimental set-up would have allowed generation and transmission of a stress-triggered chemical signal to act on leaf stomata, and exogenous ABA was indeed transmitted from roots to the shoot. However, there was no indication of such a stomata-controlling signal when we attenuated the hydraulic signal. Instead, stomata remained open in dependence of the water potential of the feeding solution.
The hydraulic signal preceding ABA must be decoded in the shoot, leading to ABA action. Complete turgor loss in leaf tissues has been suggested to stimulate ABA biosynthesis in leaves (Pierce and Raschke, 1980), but it is not known how the mechanical signal is sensed to activate ABA for stomatal closure. Hydraulic forces generated in the capillary system of plants might stimulate stretch-activated sensors (Kung, 2005; MacRobbie, 2006; Sukharev and Corey, 2004) that subsequently elicit ABA action, e.g. by enhancing ABA biosynthesis. Within the hydraulic continuum of the root system, water availability in various sectors is integrated, and the resulting hydraulic information is transmitted to the above-ground plant parts. The hydraulic mode of signal transfer allows rapid relay of information over long distances. In non-compressible pipes, the hydraulic signal can propagate at the speed of sound (Malone, 1993; Wildon et al., 1992), providing an efficient tool for communicating the water status between root and shoot to optimize plant performance and photosynthesis in a environment of limiting and changing water availability.
Plant materials and growth conditions
Arabidopsis thaliana seeds were cultivated in Petri dishes containing Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) supplemented with 10 g/l sucrose and 0.9% agar as described previously (Christmann et al., 2005). Plates were kept for 2 days at 4°C, and thereafter at 22°C under continuous illumination (60 μE m−2 sec−1). Seedlings of maple (Acer pseudoplatanus) and beech (Fagus sylvatica) were grown on MS plates without sucrose under long-day conditions with 16 h light (160 μE m−2 sec−1), at 22°C and 16°C during day and night, respectively.
Grafts of pAtHB6::LUC reporter lines (Christmann et al., 2005), wild-type (Columbia) and aba2 mutant background were generated using 5-day-old seedlings by a micrografting technique (Turnbull et al., 2003). Grafts were kept under short-day conditions with 10 h light (50 μE m−2 sec−1), 22°C, and 14 h dark, 17°C, on solidified mineral MS medium. Petri dishes with grafts were sealed with both Nescofilm (Bando Chemical; http://www.bando.co.jp; lower half) and a porous tape (Soehngen Pore; Soehngen; http://www.soehngen.de; upper half), to allow gas exchange with the chamber environment while reducing evaporation. Grafts were transferred to fresh Petri dishes every 2 weeks, and were used for experiments after development of at least eight rosette leaves (approximately 5 weeks after grafting).
ABA and stress treatments
During stress treatments, the roots of plants were selectively exposed to solidified medium such that the shoot was not in contact with the root medium. The water potential of the original root medium (MS: ψ = −0.2 MPa) was adjusted by adding mannitol. The osmotic stress medium (OSM) contained 0.32 m mannitol in MS (OSM: ψ = −1.0 MPa). Seedlings (4 days old) and grafts (5–6 weeks old) of Arabidopsis, seedlings of maple (A. pseudo-platanus, 3 weeks old), and beech (F.sylvatica, 3 weeks old) were water-stressed for 24 h on OSM or incubated on MS as a control unless otherwise indicated. To attenuate the hydraulic signal, either water (supplemented with mannitol or not) or a buffered solution (5 mm MES in water, pH 4.8–7.2, or 5 mm Tris–Cl, pH 7.6 and 8) was applied as a droplet (3–5 μl) or repeatedly for 8 h as microdroplets (0.4 μl) to the adaxial side of leaves or cotyledons. Alternatively, the shoot was brought into contact with a solidified nutrient solution, which was physically separated by a small air gap (1 mm) from the root medium. Subsequently, stomatal apertures were determined or luciferase activity was recorded. A solution of 5 mm (+)-ABA (Lomon Bio Technology; http://www.lomonbio.com) in buffer (10 mm MES in water at pH 7) was used as a stock for feeding experiments.
Measurement of stomatal aperture
Stomatal apertures were determined in cotyledons or epidermal peels from Arabidopsis leaves, as described previously (Himmelbach et al., 2002). The lengths and widths of the stomatal pores were measured using digital photographs and imagej 1.32j software (http://rsb.info.nih.gov/ij/). The stomatal aperture is given as a ratio of the pore width vs. length (Roelfsema and Prins, 1995). Stomata with a pore width of 0.2 μm or less were considered as ‘closed’, while those with a pore width >0.2 μm were defined as ‘open’.
ABA was extracted from seedlings using methanol containing 10 pmol of [2H]-ABA per sample, and was quantified by GC-MS/MS according to the method described by Müller et al. (2002).
Solidified MS medium that had been in contact with roots of 60 seedlings for 24 h was freeze-dried after addition of [2H]-ABA standard, and extracted with methanol. Acetic acid (0.1 N) was added to the methanolic extracts, which were then reduced to the aqueous phase and loaded onto C18 solid-phase cartridges (Chromafix C18ec, Macherey-Nagel; http://www.macherey-nagel.com). The cartridges were washed with acetic acid (0.1 N) and acetic acid (0.1 N):methanol (85:15, v/v). ABA was eluted from the cartridges with acetic acid (0.1 N):methanol (50:50, v/v) and quantified by GC-MS/MS as described above.
Turgor measurements were carried out on intact Arabidopsis plants grown under well-watered conditions (−0.01 MPa soil water potential) in 0.04 m × 0.04 m pots using the cell-pressure probe technique (Steudle, 1993). We employed the instrumentation described by Henzler et al. (1999). Intact leaves attached to the plant were mounted with the help of magnetic bars on a metal plate covered by a layer of cotton wool. The cotton wool layer was either dry ambient air (AIR)–OSM treatments) or soaked with deionized water using a circulating system (AIR+H2O–OSM treatments). Leaves were incubated for 2 h prior to the start of the measurements. A mesophyll cell in the midrib area of a rosette leaf was then punctured with the cell-pressure probe and turgor pressure was recorded. Deionized water or sorbitol solutions with a water potential between –0.1 and –2.0 MPa were applied to the soil (twice the maximum water holding volume of the soil), and changes in turgor pressure were monitored until the turgor remained constant again for at least 5 min. Treatment was continued with further application of appropriate solutions of sorbitol or deionized water.
Data were analysed using the Mann–Whitney U test and winstat® software (R. Fitch Software; http://www.winstat.de). All experiments were repeated at least three times with similar results.
We are grateful to the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie for their financial support to S. Hüchering for help with establishing the grafting technique, to Dr F. Assaad for useful discussions, to U. Schuberth for help with the figures, to P. Düchting for skilful technical assistance, to Dr S. Pollmann for help with the GC-MS/MS analysis, and to Y. Kim for introducing A.C. to the cell-pressure probe technique.