Apoplastic barriers to radial oxygen loss and solute penetration: a chemical and functional comparison of the exodermis of two wetland species, Phragmites australis and Glyceria maxima


Author for correspondence: Aleš Soukup Tel: +420 221951697 Fax: +420 221951704 Email: asoukup@natur.cuni.cz


  • • Few studies have examined exodermal development in relation to the formation of barriers to both radial oxygen loss (ROL) and solute penetration along growing roots. Here, we report on the structural development, chemical composition and functional properties of the exodermis in two diverse wetland grasses, Glyceria maxima and Phragmites australis.
  • • Anatomical features, development, the biochemical composition of exodermal suberin and the penetration of apoplastic tracers and oxygen were examined.
  • • Striking interspecific differences in exodermal structure, suberin composition and quantity per unit surface area, and developmental changes along the roots were recorded. Towards the root base, ROL and periodic acid (H5IO6) penetration were virtually stopped in P. australis; in G. maxima, a tight ROL barrier restricted but did not stop H5IO6 penetration and the exodermis failed to stain with lipidic dyes. Cultivation in stagnant deep hypoxia conditions or oxygenated circulating solution affected the longitudinal pattern of ROL profiles in G. maxima but statistically significant changes in exodermal suberin composition or content were not detected.
  • • Interspecific differences in barrier performance were found to be related to hypodermal structure and probably to qualitative as well as quantitative variations in suberin composition and distribution within exodermal cell walls. Implications for root system function are discussed, and it is emphasized that sufficient spatial resolution to identify the effects of developmental changes along roots is crucial for realistic evaluation of exodermal barrier properties.


Flooded soils are often characterized by oxygen shortage and accumulations of potentially toxic reduced inorganic ions and compounds derived from anoxic microbial organic matter decomposition (Patrick & Turner, 1968; Ponnamperuma, 1984; Cizkova et al., 1999). Wetland plants, however, are adapted both metabolically and structurally in numerous ways to overcome these threats. In the majority of wetland plants, aerenchyma development raises cortical porosities to much higher values than those found in nonwetland plants (Justin & Armstrong, 1987; Seago et al., 2005), reducing respiratory demand and at the same time facilitating internal gas-phase oxygen transport to roots and rhizomes and CO2 removal to the shoot (Armstrong, 1979; Greenway et al., 2006). The consequences of this include functional dependence of rooting depths on porosity (Justin & Armstrong, 1987) in flooded soil and improved potential for radial oxygen loss (ROL) from root to sediment. The latter leads the oxidation and/or immobilization of some potential phytotoxins external to the root and some important aerobic microbial activity in the rhizosphere (Frenzel et al., 1992; Begg et al., 1994; Kirk & Kronzucker, 2005).

In wetland species, ROL is usually greatest near the root apex, declining markedly in subapical parts (Armstrong, 1979; Colmer, 2003b). Because oxygen concentrations within roots increase towards the base, any subapical decline in ROL clearly indicates the development of a barrier to oxygen release. Such a barrier was first identified in bog species and in rice (Oryza sativa) (Armstrong, 1964, 1971) and tentatively attributed to lignin and suberin deposition in or on the root surface. Subsequently the barrier has been shown to vary in strength from species to species (Laan et al., 1989; McDonald et al., 2002; Garthwaite et al., 2003) and, in rice, to be inducible by transferring roots from aerated hydroponic conditions to stagnant anaerobic agar medium (Colmer et al., 1998a) and to differ among rice varieties (Colmer, 2003a). Such a barrier on the cortex periphery might function also to restrict the intake of flooded-soil-born phytotoxins (Soukup et al., 2002; Armstrong & Armstrong, 2005). Anatomical preparations, apoplastic tracer and ion movements and oxygen microelectrode studies suggest that the main barrier to ROL in Phragmites and rice appears to be the suberized part of the hypodermal cell cylinder (the exodermis) (Armstrong & Armstrong, 2001, 2005; Soukup et al., 2002). However, the barrier to ROL in the immediately subapical regions clearly also has a respiratory component in so far as oxygen consumption within the living hypodermal cells acts synergistically with physical resistance in these regions to impede the radial efflux of oxygen from the root (Armstrong & Armstrong, 2001). For this reason, it is conceivable that the reduction of ROL might occur stronger and nearer to the root tip than would be the case for compounds not consumed during their passage.

A separate line of study has concentrated on the identification and role of the suberized cell layers that form apoplastic barriers in the control of nutrients and water uptake (Ferguson & Clarkson, 1976; Clarkson et al., 1987; Gierth et al., 1998; Zimmermann & Steudle, 1998). Recently, detailed chemical analyses of suberin and lignin cell wall deposits in the endodermis as well as the exodermis have been carried out for several species (Schreiber et al., 1999, 2005b). The only study dealing in any detail with chemical modification of hypodermal cell walls and their resistance to ROL is that of De Simone et al. (2003), where the composition and content of suberin and its aliphatic–lipidic domain in particular were related to the efficiency of the restriction of radial losses of oxygen from roots and to flooding survival of four Amazonian tree species. However, only the apical 30 mm of the root was examined, and it was clear from the oxygen microelectrode profiles that oxygen consumption in the hypodermal and epidermal layers was also contributing to the expression of the barrier.

Examples of increasing subapical resistance to nutrient and water intake have been reported frequently (e.g. Ferguson & Clarkson, 1976; Zimmerman & Steudle, 1998), but results have been variable and barriers as strong as those previously noted for ROL have rarely been recorded. As Hose et al. (2001) pointed out, however, the extent and rate at which roots suberize may depend upon stress conditions such as drought, aeration, and heavy metal or nutrient stress. In maize (Zea mays), for example, hypodermal permeability to 86Rb (rubidium) and water can be reduced by an order of magnitude (Clarkson et al., 1987) by growing roots in damp air; stagnant agar also stimulates the onset of suberization (Soukup et al., 2004; Enstone & Peterson, 2005). The reason that barriers appeared leaky or incomplete, in some cases, may have been that studies did not include the more basal parts of the roots, or lacked the resolution to discriminate sufficiently between different zones along the root. For example, in Phragmites australis roots, in mainly oxygen-impermeable parts, numerous relatively unsuberized/weakly suberized cell areas (‘windows’) can be found (Justin & Armstrong, 1987; Soukup et al., 2002) that retain some permeability to oxygen (Armstrong et al., 2000) and other tracers (Soukup et al., 2002). Lateral roots eventually emerge at these points. These have also been noted in maize (Peterson & Enstone, 1996) and rice (Armstrong & Armstrong, 2001, 2005) and it is as yet unclear to what extent they could mask the expression of barrier function in whole-root studies on nutrient and water uptake.

The current study was undertaken to examine further the relationships between the pattern and composition of suberin deposits within exodermal cell walls and the ROL and solute permeability properties in adventitious roots. For this purpose, two contrasting wetland grasses, Phragmites australis and Glyceria maxima, were chosen. Both species are rhizomatous but they differ in ecophysiological preferences (Buttery & Lambert, 1965; Buttery et al., 1965) and growth strategy. G. maxima produces a superficial root system in shallow flooded zones while P. australis colonizes habitats from shallow to very deep water (Buttery & Lambert, 1965; Hejny & Husak, 1978). Convective gas flows through the rhizome system from the shoots occur in P. australis (Armstrong et al., 1992) but have not been found in G. maxima (Vretare Strand, 2002), while P. australis rhizome tissues are more tolerant to strict anoxia (Braendle & Crawford, 1987). Vlkováet al. (2004) showed, in solution culture experiments, that the two species differ significantly in accumulation of toxic metals from cultivation solution, an observation that suggested some uncontrolled penetration of the solutes into root tissues, and perhaps differences in exodermal function influencing the ecophysiological tolerance of these species.

Materials and Methods

Plant cultivation

Plants of Phragmites australis (Trin.) ex. Steud. and Glyceria maxima (Hartm.) Holmb were propagated from rhizome cuttings collected from stock plants obtained from the Institute of Botany in Trebon (South Bohemia). Cuttings bearing up to three stalks had their existing roots trimmed back before being placed in a quarter-strength Hoagland III (Hoagland & Amon, 1950) solution containing the following micronutrients: H3BO3, 0.23 µm; MnCl2, 0.71 µm; ZnSO4, 5 nm; (NH4)6Mo7O24, 0.6 nm; CuSO4, 1.6 nm. The nutrient solution was either slowly circulating between the cultivation vessel (15-l container) and a reservoir (40 l) or was kept stagnant (in the 15-l container). There were six plants to each of two containers per treatment. The circulating solution was well oxygenated (6–7 mg O2 l−1) during mixing of the solution in the reservoir. Before use, the stagnant solution (max. 0.15 mg O2 l−1) was enriched with 0.05% agar and deoxygenated at least overnight by flushing with nitrogen. Both solutions were changed regularly every 7th day. Plants were grown for 7 or 14 wk in a cultivation room (day:night 16 : 8 h; 20 : 16°C; 40 : 60% relative humidity) before sampling.


Segments from subapical and basal parts (within 2 cm from the root–shoot junction) of new roots were sampled, sectioned with the aid of a hand microtome and stained with sudan red 7B or fluorol yellow (Brundrett et al., 1988b) to visualize lipidic compounds, HCl-phloroglucinol (Johansen, 1940) to visualize lignin compounds and berberine-toludine blue O (Brundrett et al., 1988a) to visualize the Casparian bands and suberin lamellae in either ultraviolet (UV) or blue light excitation. Berberine was used as a 0.1% solution for 1 h and counterstained with 0.5% toluidine blue for 10 min. The sections were afterwards mounted in 50% glycerol. The same procedure was used after extraction with a CHCl3/CH3OH mixture, which was expected to ‘unmask’ lipids (Gahan, 1984). The plasmalemma was stained with the FM4-64 probe (Molecular Probes, Eugene, OR, USA) to observe plasmolysis induced with 2 m sucrose. The permeability of hypodermal layers was assessed using periodic acid (0.1% aqueous solution) as described in Soukup et al. (2002). The roots were subsequently washed in water and the remaining H5IO6 was removed in the following reducing solution: 1 g of KI and 1 g of Na2S2O3·5H2O dissolved in 50 ml of H2O and acidified with 0.5 ml of 2 m HCl. Aldehydes produced in cell walls by H5IO6 were detected in sections with Schiff's solution (for details, see Pearse, 1968). The possible toxic effect of H5IO6 on protoplasts during long-term exposure was taken into consideration when interpreting the results.

Hypodermis suberin analysis

Root segments for suberin analyses were harvested from two independent cultivations. In the first cultivation, after 14 wk, basal segments 65 mm long were taken from well-established roots 216–446 mm long. In the second cultivation, after 7 wk, segments 0–30, 30–70 and 70–110 mm behind the root tips were sampled from growing roots (length 85–155 mm). Roots of similar total length and with apical regions of comparable length without laterals were grouped into single samples (usually three to five roots per plant) in order to obtain enough material for biochemical analyses. At least five replicates were included for each position and treatment, with the exception of segments at 70–110 mm, where only three replicates were evaluated because of small amounts of material. Sampled segments were blotted, weighed and scanned at 600 dots per inch (dpi) and their surface areas were determined using the ImageJ analyser (http://rsb.info.nih.gov/ij). The dry weight was determined after drying at 65°C for 3 d. Hypodermal and attached epidermal cell layers were enzymatically isolated in a mixture of cellulase (0.5% weight/volume (w/v); Fluka, Buchs, Switzerland) and pectinase (0.5% w/v; Fluka) in citrate buffer (pH = 3, 100 mm) as described by Schreiber et al. (1994).

Isolated hypodermal material (0.5 mg) was trans-esterified with BF3 and the released material processed as in Zeier & Schreiber (1998). For quantification and identification of the released monomers, we used an HP 5890 Series II gas chromatograph (Hewlett Packard, Palo Alto, CA, USA), equipped with a flame ionization detector, and an Agilent 6890 N gas chromatograph (Agilent Technologies, Böblingen, Germany) combined with a quadropole mass selective detector, Agilent 5973 N. The content of suberin monomers was expressed per unit of root surface area instead of per unit dry weight of hypodermal material, as this proved the most reliable and stable measure of quantity within the root wall and also the most physiologically relevant.

Radial oxygen loss from root

Radial oxygen loss (ROL) was measured along adventitious roots using sleeving cylindrical platinum cathodes in conjunction with Ag-AgCl anodes, after the method of Armstrong & Wright (1975), in freshly deoxygenated 0.05% w/v aqueous agar containing quarter-strength Hoagland's solution. ROL profiles were taken along the root axis from ∼3 mm behind the root tip up to emerging laterals and slightly beyond, as far as the laterals were not too long or too numerous to be injured by dragging the electrode along the root. Radial profiles of oxygen concentration were measured with microelectrodes as described in Armstrong et al. (2000).


Results of biochemical analyses were statistically evaluated by analysis of variance (ANOVA), t-test, paired t-test to reveal differences in one or more discrete variables, and the Armitage test for trends in proportions (NCSS, 2000; ©Jerry Hintze, Kaysville, UT, USA). Sigmaplot 9 (Systat Software, Inc., Erkrath, Germany) was used for drawing graphs and regression analysis.



The anatomical part of the study focused on the root cortex. Cells of the middle cortex of the apical parts of both species are radially aligned, arranged in a cubic pattern (Justin & Armstrong, 1987), and subsequently develop similar forms of lysigenous aerenchymatous channels. The difference between the species is most pronounced in the outer cortex, with each having a distinctive form of hypodermis (Fig. 1). This develops in both species in a constitutive way and contains living protoplasts (data not shown) even in the basal parts of the roots; it is also present in well-aerated solution or soil.

Figure 1.

Comparison of the structure of the hypodermis and the endodermis at the bases of well-established roots of Phragmites australis (a–c) and Glyceria maxima (d–e). Bars, 50 µm. (a) HCl-phloroglucinol detection of lignin in the hypodermis (hyp) of P. australis and thickened cell walls of its inner sclerenchymatous ring (scr); (b) suberized exodermis (exo) stained with sudan red 7B; (c) sudan-stained suberin lamellae of the endodermis; (d) lignin or lignin-like compounds were detected with HCl-phloroglucinol (red colouring) in hypodermal layers (hyp) of G. maxima with typically thickened middle tangential cell walls (arrow); epi, epidermis; (e) a section comparable to (d) was stained with sudan red; (g) suberin lamellae (arrow) of the endodermis stained with sudan red from the same section as (e).

The hypodermis of P. australis is composed of a uniform multilayered exodermis and underlying nonexodermal sclerenchymatous ring (Fig. 1a,b). Discernible Casparian bands appeared in the exodermis but were only visible for a short period during differentiation (for a detailed description of exodermis development, see Soukup et al., 2002), being later obscured by complete suberin lamellae. Complete suberin lamellae of the exodermis were, similarly to the endodermis (Fig. 1c), strongly stained with sudan red 7B (Fig. 1b) and fluorol yellow. Suberization was not histochemically detected in the sclerenchymatous ring of the hypodermis (Fig. 1b). Hypodermal cell walls were positive for lignin using HCl-phloroglucinol (Fig. 1a).

G. maxima also possesses a distinct uniform hypodermis, but its character differs from that of P. australis. Its exodermis consists of two cell layers with markedly thickened adjacent tangential cell walls (Fig. 1d,e). Sometimes one subhypodermal layer may additionally develop thickened sclerenchymatous cell walls and appear to belong to the hypodermis, but fine intercellular spaces are present in the corners of cell walls and therefore we considered it to be part of the middle cortex. Further thickening of hypodermal and underlying layers may occur in some roots. Early modifications of the exodermal cell walls were recognized in G. maxima as a deposition of autofluorescent material. This significant increase in autofluorescence of exodermal cell walls was discernible very close to the root tip (up to 10 mm) and was even closer than similar cell wall modifications of the endodermis. Suberization of G. maxima hypodermis was never detected with sudan red (Fig. 1e) or fluorol yellow staining. This failure to detect suberization in the hypodermis was not caused by a deficient staining procedure as intense staining of the endodermis within the same section provided a reliable positive control (Fig. 1f). The negative sudan staining did not improve after CHCl3/CH3OH extraction, which might unmask some lipids, as suggested by (Gahan, 1984). In spite of the lack of staining for suberin, Casparian bands were discernible for a considerable distance behind the root tip, being detected with berberine-toluidine blue staining (e.g. at 7 cm in Fig. 2a) and by band plasmolysis of the exodermis where plasmalemmae were closely adhering to cell walls in regions of Casparian bands (Fig. 2b). Casparian band staining and band plasmolysis could not be detected in older root parts after deposition of additional cell wall material in the hypodermis (Figs 2c–f). The location of suberin impregnation of hypodermal cell walls can also be estimated from a comparison of the autofluorescence of unstained sections (Fig. 2c) with that of similar sections counterstained with toluidine blue (Fig. 2f) from basal parts of well-established roots. Toluidine blue quenches cell wall autofluorescence if it is bound. The presence of suberization might restrict toluidine blue penetration and binding, and thus prevent possibly quenching of cell wall autofluorescence. The suberization is thus indicated as a complete lamella in the inner cell layer and incomplete lamellae on inner periclinal and anticlinal cell walls of the outer exodermal layer. The most obvious evidence of suberization comes from the thickened adjacent tangential cell walls. Particularly in this location, periclinal striation of the fluorescence was observed in well-developed parts of the root (Fig. 2d), which seems to indicate that several suberin lamellae are present in this region of the cell wall. The location of suberization within the hypodermis is supported by results of sulphuric acid digestion (Fig. 2g), as only suberin is considered to resist digestion (Johansen, 1940).

Figure 2.

(a) Casparian bands (arrows) in Glyceria maxima exodermis stained with berberine-toluidine blue in basal parts of a root 150 mm long bearing laterals emerging at 105 mm behind the root tip; epi, epidermis; bar, 50 µm. (b) Band plasmolysis (arrows) of the protoplast with plasmalemma attached to Casparian bands of the exodermis. Membranes were stained with FM4-64 and plasmolysed in sucrose solution. The root base of a 70-mm-long root with no laterals is shown; bar, 50 µm. (c, d) Cross-sections from the basal part of a well-established 450-mm-long root. (c) Autofluorescence under ultraviolet (UV) excitation. The cells of the epidermis (epi) were dead and sloughing. (d) Detail of striation of tangential thickened exodermal cell wall from (c). (e) Berberine-toluidine blue staining under blue excitation; arrows point to thickened tangential walls of hypodermal layers. (f) A toluidine blue-stained section of the base of a 230-mm-long root, showing autofluorescence under blue excitation. (g) The exodermis (arrows) digested with concentrated sulphuric acid; bars, 50 µm.

Chemical analyses of suberin of hypodermal layers

Suberin monomers were detected in the hypodermal layers of both species. Similarly to previous work, epidermal cell walls and lignified nonexodermal sclerenchymatous cell walls were included in the detection process as they could not be enzymatically separated from exodermal layers (De Simone et al., 2003; Schreiber et al., 2005b). Epidermal cells formed only a minor fraction of any isolated sleeve of cell walls or were even sloughed off in the basal parts of roots. As with the partially included sclerenchymatous ring of P. australis, they also did not respond to histochemical lipid tests and dissolved in sulphuric acid. Therefore, we conclude that they introduced only minor or no distortion of the results of biochemical analyses of the suberin content of the hypodermis. Considerable variation in the thickening of the hypodermal cell walls was recorded, which was inconsistent between cultivations as well as between treatments for both species. It seems that other environmental factors (movement of the nutrient solution, the mechanical properties of the rooting medium, etc.) might be more important for development of this feature than oxygen concentration alone. Thickening of the hypodermal cell walls affected values of dry weight per unit root surface area of total enzymatically extracted hypodermal material, which included other compounds in addition to suberin, but did not have a pronounced effect on suberin content per unit root surface area. The total content of aliphatics per unit surface area was several times higher in corresponding regions of P. australis roots than in G. maxima (paired t-test, P = 0.001). The overall higher variation in the content of aliphatic compounds in P. australis might be ascribed to variability in the number of hypodermal layers; in G. maxima there were regularly only two.

A lower proportion of aliphatic to aromatic compounds (mean = 4.4×, standard deviation (SD) = 0.96) in G. maxima compared to P. australis was recorded at each position along the root (paired t-test, P < 0.0001, n = 8; Fig. 3). The major increase of phenolics in the basal part of G. maxima (Fig. 3) was connected with thickening of hypodermal cell walls, which are rich in simple phenolics in grass species. As mentioned in the previous paragraph, variation in the thickening of hypodermal cell walls and their lignification in G. maxima seems to be the reason for the higher variability in phenolic content in comparison to aliphatic compounds. This also suggests that not all the phenolics extracted after trans-esterification are necessarily related to the suberin aromatic domain.

Figure 3.

Contents of major suberin monomers of exodermal cell walls and their changes along the axis of roots grown in stagnant nutrient solution. The samples of basal parts correspond to those analysed in Table S1 (Supplementary Material). Error bars represent standard deviations. At least five replicates were included. The position along the root (in mm from the root tip) is indicated in the key.

Towards the bases of roots of both species, the aliphatic content per unit root surface area increased with differentiation of hypodermal tissues (Fig. 3). Also, for both species, significant increases were recorded in the rapidly developing distal parts of the root in the segments at 0–30 and 30–70 mm. Within 70 mm behind the root tip aliphatics had reached > 50% of the maximal content of the basal parts of the roots. However, the species differed significantly in relation to the composition of the aliphatic domain of hypodermal suberin (Fig. 3; for the detailed monomeric composition, see supplementary Table S1, available online). A very low content of unsubstituted monocarboxylic fatty acids (t-test, P < 0.001) was recorded in G. maxima together with higher proportions of long-chain alcohols (t-test, P < 0.001) compared with P. australis (Fig. 3). Dicarboxylic acids were detected in P. australis but not in G. maxima.

Significant changes also appeared in the proportions of individual monomers along the developing root (Fig. 3). There was an increase in the chain length of aliphatic compounds deposited into the exodermis in the more differentiated part of roots further from the root tip (Armitage test, P < 0.05; Fig. 4). This shift towards longer chain aliphatic production towards the base was more pronounced in G. maxima (Fig. 4).

Figure 4.

Relative proportions of particular chain lengths of fatty acids of exodermal suberin and their changes along developing roots grown in stagnant nutrient solution. At least five replicates were included per position. The position along the root (in mm from the root tip) is indicated in the key.

No statistically significant difference in either composition or content of analysed compounds was found to be connected with cultivation in stagnant and circulating solutions.

Permeability of the exodermis

The exodermis acted as a peripheral apoplastic barrier on the root periphery in both species. Penetration of apoplastic tracers as well as ROL from the roots were restricted as differentiation of hypodermal cell walls proceeded.

The profiles of ROL along G. maxima roots from stagnant and nonstagnant solutions are shown in Fig. 5 (for individual profiles, see Fig. S1 in the supplementary material). In roots from both conditions, ROL decreased within the first 30 mm from the root tip. Increased oxygen leakage proximal to this zone was spatially connected to emergence of lateral roots. However, this increase, which was significant in nonstagnant cultivation, was only temporary in roots after 3 d in stagnant solution and measurable ROL declined almost to zero (Fig. 5a,b), even in the region where laterals had already emerged and the electrode was carefully pulled over them. In both stagnant and nonstagnant solutions, Afreen-Zobayed (1996) recorded profiles along roots of P. australis similar in pattern to those of the G. maxima roots from the stagnant solution, although the minimum values were reached further from the tip than in G. maxima (at c. 50–60 mm rather than 30 mm). In stagnant deoxygenated medium, the minimum values were close to zero; in the aerated medium, they were c. 8% of the apical value but measurements were only possible to 85 mm, where total root length was 105 mm. In G. maxima roots grown in nonstagnant solution, ROL declined with distance from the root tip but then increased towards the base and did not decrease again within the measurable part of the root (Fig. 5). Linear regression of data from 25 mm towards the base revealed an increase (slope = 1.4; r = 0.779) for roots grown in nonstagnant solution but no significant trend (slope = −0.07; r = 0.204) after 3 d in stagnant solution.

Figure 5.

Longitudinal profiles of radial oxygen loss (ROL) measured along roots of Glyceria maxima. (a) Roots grown in nonstagnant (more oxygenated) solution (n = 4). (b) Roots grown for 3 d in stagnant hypoxic solution; high variability in the apical part reflects variation in the length of this region among individual roots (n = 5). Profiles of ROL measured for individual roots are presented in Fig. S1 (Supplementary material). Error bars represent standard deviation.

The location of the ROL barrier within G. maxima roots was studied using microelectrodes that penetrated the root wall. Measurements were taken 5 and 29 mm behind the root tip. Unfortunately, the thickening of hypodermal layers made penetration by microelectrodes difficult, and further behind the root tip penetration was impossible. Therefore, only two successful replicates were recorded at 29 mm. At 5 mm from the tip, the root wall was still quite permeable to oxygen and its leakage increased the concentration of oxygen in the surrounding anoxic medium (Fig. 6a). Enhanced barrier properties were recorded 29 mm from the root tip, where the second track was taken. In this profile, the oxygen concentration approached zero at the root surface. The inflection point of a logistic curve fitted to the recorded data showed that in both replicates the highest increase was located at c. 46 µm depth from the root surface and the major differences were thus recorded across the inner hypodermal layer (Fig. 6b). The location of the ROL barrier in P. australis roots was, similar to G. maxima, in an exodermal position (Armstrong et al., 2000).

Figure 6.

Radial profiles of oxygen concentration measured with microelectrodes 5 and 29 mm behind the tip of Glyceria maxima roots grown in stagnant solution. The profiles of oxygen concentrations correspond to the root cross-sections shown in the figure. The brown track induced by wounding indicates the path of microelectrode penetration (red arrow). Both sections were stained with HCl-phloroglucinol for lignin. A logistic curve was fitted to discrete data.

The permeability of hypodermal layers was also assessed by penetration of a solution of periodic acid. In both species, the penetration from the medium bathing the roots was greatly reduced with exodermis development (Fig. 7a,b). For G. maxima treated for 1 h, the solution pervaded the epidermis and outer tangential and partially outer radial cell walls of the outermost hypodermal layer in well-developed root parts (Fig. 7a). With more prolonged treatment (4 h), penetration was deeper (Fig. 7d) and both exodermal cell layers and adjacent cortical layers were stained. The tracer passed through the exodermis. In P. australis, only minor differences were found between the 1-h and 4-h treatments (Fig. 7c). Some additional outer exodermal cells were stained, but the deeper penetration was not regular and tracer never crossed the exodermis. The location of the cell layers that restrict penetration of periodic acid through the hypodermis, and differences in the arrangement of these cell layers, were obvious if the tracer was vacuum-infiltrated into root segments and the aerenchymatous channels of the middle cortex and then allowed to diffuse into neighbouring tissues, thus penetrating the hypodermis from both sides (Fig. 7e,f). In P. australis, the penetration stopped on the inner and outer borders of the suberized exodermis (Fig. 7e). Thickened and lignified cell walls of underlying sclerenchyma were easily penetrated (Fig. 7e). In G. maxima, the faces penetrating from the inner and outer sides met within the hypodermis and both layers were stained (Fig. 7f). Thus, the hypodermis was fully penetrated. The weaker reaction, or even lack of reaction, in the middle part of the tangential cell walls of two meeting hypodermal layers is interesting (Fig. 7f). Polysaccharides in these parts of the cell walls were obviously not oxidized with periodic acid, in spite of the fact that the acid penetrated the exodermal cell walls. The failure of the reaction might be attributable either to the absence of oxidizable material (polysaccharides) or, more likely, to the polysaccharidic material being masked by other compounds (e.g. suberin).

Figure 7.

Cross-sections of the hypodermis of Glyceria maxima (a, b, d, f) and Phragmites australis (c, e) root segments with hypodermis tested for permeability to periodic acid. (a) Periodic acid (as well as reducing solution) was applied to the surface of the basal part of a well-developed G. maxima root for 1 h. Red staining indicates the penetration of periodic acid, photographed as fluorescence under blue excitation. (b) Fully penetrated apical part (5 mm from the tip) of a G. maxima root after 1 h of external treatment. The purple coloration indicates tissues penetrated with periodic acid. (c, d) Tracer solution was applied externally to the root surface for 4 h. (e, f) Segments of length 1 cm were vacuum-infiltrated with tracing solution and left for 4 h to allow penetration. Segments (a, c, d) were sampled from the basal parts of well-established roots with length > 270 mm. epi, epidermis. Arrows indicate the middle part of adjacent cell walls with a weaker reaction. Bars, 50 µm.


Anatomical features

Both the structure and the chemical composition of the hypodermal cell layers of roots were examined in G. maxima and P. australis, monocotyledonous wetland species of Poaceae. Both species possess an exodermal hypodermis with suberized cell walls as defined by Haberlandt (1918) and von Guttenberg (1968). The P. australis hypodermis is composed of a sclerenchymatous ring and multilayered exodermis, while in G. maxima a two-layered exodermis is present (Fig. 1). The exodermal layers of P. australis develop suberin lamellae very early, and Casparian bands are obscured close to the root tip (Soukup et al., 2002). Instant or almost instant deposition of suberin lamellae over Casparian bands seems to be a general feature of the exodermis, having also been found in other species tested to date (Peterson & Perumalla, 1984; Seago et al., 1999; Ma & Peterson, 2000). In roots of G. maxima, however, in the inner hypodermal layer, the Casparian bands and band plasmolysis remained discernible for a considerable distance behind the root tip (e.g. Fig. 2b). Later formation of complete suberin lamellae and additional cell wall material deposition released the attached plasmalemma from the cell wall of the inner exodermal layer. The outer exodermal layer seems to be impregnated mostly in its inner tangential and partially radial cell walls. The heaviest deposition of suberin was in the adjacent tangential cell walls of the hypodermal layers, and several parallel lamellae seem to be formed in this part of the cell wall. The pattern thus differed from the regular multilayered lamellae of P. australis.

In most work published to date, the localization of suberin in tissue was confirmed by staining with sudan red or fluorol yellow lipophilic dyes. However, both stains failed to detect G. maxima exodermal lipidic cell wall material, the presence of which was confirmed biochemically. The staining of the endodermis can be seen as a positive control, and the striking difference in staining demonstrates the different properties of suberized cell walls not only between species but also between the endodermal and exodermal layers of G. maxima roots (Fig. 1e,f). The limitation of histochemical detection has been noted also by De Simone et al. (2003), and was interpreted as being attributable to low amounts of suberization. However, in the current case the quality and not the quantity of suberin is the main difference between the species, as even in the very apical tissues of P. australis suberin lamellae are stainable, but they are not stainable in the basal parts of G. maxima. We can only hypothesize that differences in the composition and/or molecular and spatial arrangements of suberin within the cell wall or differences in its molecular context in the cell wall may be factors affecting lipid staining. Obviously, negative results for sudan staining should not always be regarded as decisive.

Suberin of hypodermal layers

The presence of a lipophilic (aliphatic) domain of suberin is generally connected with barrier properties (Vogt et al., 1983; Schreiber et al., 1999) described for the periderm (Groh et al., 2002), and the endodermis and exodermis (Zimmermann & Steudle, 1998; Barrowclough et al., 2000; De Simone et al., 2003). The monomeric composition of suberin significantly differed between the two species (Fig. 3, Supplementary Table S1). The ω-OH alkanoic and 2-OH alkanoic acids were the most abundant aliphatic monomers in our species. Interestingly, 2-OH alcanoic acids have been reported to be a significant monomer of suberin in maize (Zeier et al., 1999b) and, in the endodermal Casparian bands, in Clivia minimata but not in other species tested to date (Schreiber et al., 1999; Zeier et al., 1999a; Bernards, 2002). No dioic acids were detected in the hypodermal cell walls of G. maxima but were present in P. australis. α,ω-dioic acid is considered to be an aliphatic compound typical of, and unique to, suberin and was even suggested as a marker to differentiate between suberized and cutinized tissue (Matzke & Riederer, 1991). The only other known exception, in which dioic acids are absent, is the suberin of endodermal Casparian strips in C. minimata (Zeier & Schreiber, 1997), which remain in the first developmental stage, i.e. without complete suberin lamellae (Clarkson & Robards, 1975). Changes in suberin monomer composition were also found to accompany differentiation of the exodermis along the growing root. Lengthening of the aliphatic chains of major monomers was recorded for both genera. Suberin deposits with higher proportions of longer chain aliphatic monomers in older parts of the root (Fig. 4), reported also by Zeier et al. (1999b) for different developmental stages of the endodermis, thus seem to be a more general feature of suberized tissues. Another developmentally related trend, recorded only for P. australis, was an increase in the relative proportions of unsaturated fatty acids towards the base. The above-mentioned developmental changes in the composition as well as in the content of suberin were not restricted to particular developmental stages of the exodermis or to the short subapical region (‘differentiation zone’) of the root, but took place along considerable lengths of the root behind the root tip in both species.

The content of aliphatic suberin in hypodermal layers of both P. australis and G. maxima reached values comparable in order of magnitude to values measured for other plant species (for comparison, see De Simone et al., 2003; Schreiber et al., 2005a). However, considerable variation exists among tested species. In the case of Amazonian trees (De Simone et al., 2003), suberin contents measured in the first 30 mm were comparable to, or even higher than, those measured in the basal parts of the roots of our species. The more than three times higher content of aliphatics in the P. australis exodermis compared with G. maxima seems to be connected not only to a higher concentration within the cell wall material but also to a different pattern of suberin deposition and a different number of suberized exodermal cylinders. We did not find any significant difference in overall suberin content and composition between roots grown in stagnant and circulating nutrient solutions. This conclusion might be related to constitutive development of the exodermis in both species.

Barrier properties of the exodermis

The barrier properties of the exodermis have been described in the literature dealing with water transport (Ranathunge et al., 2003), ion uptake and penetration (Gierth et al., 1998), apoplastic tracers (Peterson et al., 1982; Soukup et al., 2002) and movement of plant growth regulators (Seago et al., 1999; Hose et al., 2000). The properties and development of hypodermal layers are strongly affected by environmental conditions in some species (Enstone & Peterson, 1998; Degenhardt & Gimmler, 2000; Hose et al., 2001; Soukup et al., 2004) and affect the communication of root tissues with the near rhizosphere. Barriers formed subapically in hypodermal layers of wetland plants or plants from flooded soil might, by restricting ROL, be beneficial for longitudinal oxygen transport in roots (Armstrong, 1979). They can also regulate the passive transport into and out of the root of other compounds from solution in close contact with the roots. In most wetland plants tested to date, the proportion of tissues that are not shielded from the often harsh rhizosphere conditions within flooded soil seems to be minimized by the very early development of the exodermis (Seago et al., 1999, 2000a,b). Modifications of exodermal layers even precede modifications of the endodermis and appear closer to the root tip in P. australis (Soukup et al., 2002) and G. maxima. The inverse developmental sequence has been recorded for nonwetland plants, where the exodermis develops some considerable distance behind the root tip (Ferguson & Clarkson, 1976; Perumalla & Peterson, 1986). However, even in nonwetland plants the very early differentiation of the exodermis can be induced in some plants as a consequence of cultivation in rooting medium low in oxygen (Soukup et al., 2004; Enstone & Peterson, 2005) and reversed when plants are transferred into well-oxygenated soil (Soukup et al., 2004). This seems not to be the case for G. maxima and P. australis roots, where the exodermis developed close to the tip as a constitutive feature regardless of oxygen availability and also in well-aerated soil (Soukup et al., 2002).

In the literature, the restriction of ROL and barrier formation have been linked to various features of tissues external to the aerenchyma. The dense arrangement of the hypodermal cylinder (Connell et al., 1999), thickening of hypodermal cell walls (Colmer et al., 1998b; Armstrong et al., 2000) and suberization and/or lignification (Armstrong et al., 1994a) were suggested to have these functions. In both species tested in our work, the barriers were shown, using microelectrodes (for P. australis, see Armstrong et al., 2000), to be in the suberized exodermis. As no other observed anatomical features other than suberin impregnation of the hypodermal cell wall were recorded in the short region (at c. 30 mm) of sharp ROL decrease along the root, the key role of this marked increase in setting up barrier properties seems obvious. The tracing of the location of the barrier to the suberized exodermis and not the thickened and lignified cell walls of sclerenchymatous hypodermis was further supported by periodic acid penetration tests. The thickened and lignified hypodermal cell walls of the sclerenchymatous ring of P. australis did not provide a significant barrier to diffusion of periodic acid and is also unlikely to be a barrier to diffusion of smaller molecules such as oxygen. Recently, De Simone et al. (2003) reported a correlation between suberin content and improved survival of tropical trees in flooded soil, thus confirming experimentally for the first time the connection between suberization and the barrier to ROL in roots. Our data provide further, more direct evidence supporting this conclusion.

Major differences between the ROL profile patterns along roots grown in stagnant and nonstagnant solutions were found in G. maxima, but not in P. australis (Afreen-Zobayed, 1996). The absence of statistically significant differences in suberin content and the measured initial reduction in ROL of the subapical region of roots from both treatments show that even in nonstagnant cultivation some barrier is formed in the apical region. Reasons for the basipetal increase in ROL might include some of the following. (a) The emerging laterals puncture the exodermis and open up a radial pathway across the epidermis and hypodermal layers. We can only speculate as to whether, for example, the difference in the effectiveness of resealing after the emergence of laterals might result in a difference in the spatial pattern of ROL among treatments. (b) The ROL from laterals developed in nonstagnant medium might be higher than that from laterals developed in stagnant medium. (c) The creation of the ROL barrier in the near subapical region, which is the most metabolically active part of the root, was so quick in G. maxima that it took place during measurement (within several hours) in stagnant medium, but the barrier did not form in the more differentiated distal parts of the root. The evaluation of the functional properties of the root exodermis might be affected by cultivation treatments and/or by disregarding the presence of gaps in otherwise impermeable parts of the root, which are associated with the pre-emergent stage of lateral root development, as seen in this and other studies (Armstrong et al., 2000; Soukup et al., 2002). Such interruptions are below the spatial resolution of most measurement techniques of root radial hydraulic conductivity, solute penetration and ROL, which are almost exclusively performed on those parts of the roots still bearing no laterals. These factors therefore should be taken into account.

Although suberization sufficient to create a ROL barrier appeared in the immediately subapical regions of the roots of both species, the efficiency of the exodermis in oxygen retention might be attributable in part to synergism between respiratory demand and radial diffusive resistance (Armstrong et al., 2000), an effect magnified in apical parts by high respiration rates. However, suberin content further significantly increased in the basipetal direction. A correlation between the total amount of suberin in hypodermal layers and the ability of the plant to withstand flooding (De Simone et al., 2003) thus might not always necessarily reflect only the oxygen barrier properties of the root, as the suberin content necessary to form ROL barrier is several times exceeded at some distance from the root tip, at least in the species tested. The correlation might be also related to other environmental factors of flooded soil. One of these factors might be the presence of phytotoxic compounds, which are characteristic of flooded soils (Ponnamperuma, 1984). The periodic acid penetration into root tissues across the exodermis illustrates this function of the exodermis and interspecific differences. In comparison with the tight barrier in P. australis, the penetration of the exodermis of G. maxima by periodic acid may indicate a lower selectivity of passive uptake via roots from the rhizosphere in G. maxima. Such a difference could explain the approximately three times higher accumulation of aluminium from the rooting medium in root tissues and the greater degree of middle cortex injury in G. maxima compared with P. australis (Vlkováet al., 2003). However, the causal connection requires further evidence.

In addition to interspecific differences in exodermis structure, composition and function, these comparisons demonstrate the variation in barrier permeability to different substances. Studies on other species have indicated that water intake is not prevented by exodermal suberization, for example in rice (Ranathunge et al., 2005; Schreiber et al., 2005b), and appears to contrast with tight ROL barrier properties (Colmer et al., 1998, 2003a). There may be several reasons for this. One possible reason is that the aerated hydroponics used in most studies (e.g. Ranathunge et al., 2005; Schreiber et al., 2005b) did not, in the species used, induce ROL barriers as strong as those developed under wetland conditions or in deeply hypoxic stagnant media. Another may relate to the positions along the root that were measured; the work on hydraulic conductivity has been neglectful of more basal regions. Recently, Garthwaite et al. (2006) observed that the induction of a ROL barrier with a stagnant nutrient solution did not significantly affect the hydraulic conductivity of Hordeum marinum roots, but the barrier to ROL was not as complete as those found here or previously in rice in stagnant anaerobic media. In contrast, Groh et al. (2002) demonstrated that the diffusional permeance of oxygen in suberized phellem of several woody species was at least an order of magnitude lower than that for water. The mechanism underlying the difference in permeability is not clear, but molecular size may play a role in transport. The water molecule, with a mean van der Waal diameter of c. 280 pm (Finney, 2001), might permeate the apoplast more easily than the oxygen molecule, with a van der Waal diameter estimated at 428 pm.

Schreiber et al. (2005b) suggested that gross values of suberin content in hypodermal layers do not necessarily reflect the barrier properties of impregnated cell walls, and this suggestion is supported by the current data. The spatial arrangement of suberin deposition within the cell wall and cylinder of the exodermis should be further studied at sufficient resolution to provide the data required to elucidate the transport and pathways of various solutes across such apoplastic barriers and to explain the reported interspecific differences.


The presented work was supported by grant GAUK 274/2004/B-Bio/PrF. The authors are very grateful to the referees for their valuable comments.