Total element concentrations and uptake by Thlaspi caerulescens
Thlaspi caerulescens always had a higher biomass when grown in Giornico soil (acidic) than in Dornach soil (calcareous) although this species is usually defined as a calcareous soil indicator ( Lauber & Wagner 1996). Combined with higher Zn concentrations, which were also present in plants, the resulting Zn uptake increased when the Zn bioavailable pool was larger, as was found by Brown et al. (1994) . Indeed, the uptake was not related to total (HNO3-extractable) Zn concentrations in T. caerulescens; the total quantities of Zn extracted by this plant from the Giornico soil were twice as high as those removed from the Dornach soil, although the total (HNO3-extractable) Zn concentrations in the Giornico soil were only half those in the Dornach soil. As a consequence, Thlaspi caerulescens removed 10% of the initial total Zn in the Giornico soil in 24 weeks instead of less than 2% in the Dornach soil. This amount was larger than the amount predicted by extraction with NaNO3; T. caerulescens was able to remove only twice the equivalent of the NaNO3-extractable Zn present in the soil of Giornico, whereas in Dornach soil it could extract the equivalent of 100 times the NaNO3-extractable Zn. Thus, even though the plants were grown in pots, that is in a confined environment which might have enhanced the extractability, this result raises the question of the significance of NaNO3 extraction as a tool to assess metal availability to T. caerulescens ( Gupta & Aten 1993) and, as a consequence, the type of metal pool which is depleted by T. caerulescens. Brown et al. (1994) suggested that when Zn concentrations in soil solutions are limiting (which is the case for Dornach), an active mechanism could be responsible for the discrepancy between uptake and indices of Zn bioavailability in the soil. Such a mechanism may occur in the Dornach soil, in which NaNO3-extractable Zn is very low.
Cadmium concentrations in the T. caerulescens tissue were always very low, although T. caerulescens has also been reported to hyperaccumulate Cd ( Robinson et al. 1998 ). These relatively low Cd concentrations are related to the soil concentrations, which were below 19 μmol kg−1 of HNO3-extractable Cd. The Cd concentrations, however, were in the same range as that found by Brown et al. (1995b) originating from soils with similar Cd concentrations. The relatively low Cd concentrations might also be related to the Thlaspi genotype itself ( Meerts & Van Isacker 1997).
At the plant level, the concentrations of Zn and other elements in T. caerulescens grown on both Dornach and Giornico soils showed changes with time. In the shoots, all elements except Fe and Cu became diluted with time, whereas in the roots some elements such as Zn, Ca, Mg, Fe, Cu and Cd accumulated. A similar decrease over time has also been described by Brown et al. (1995b) for Mg and Ca in leaves of T. caerulescens grown in sludge-amended soil, and by Rascio (1977) in field conditions. However, in pot experiments lasting 6 months, Meerts & Van Isacker (1997) measured concentrations of major elements which were in general higher than those found by Brown et al. (1995b) as well as higher than those detected in the present experiment. These differences may be related to variations in plant populations, age of the plants or decrease in supply of bioavailable elements to the plants.
The Zn concentrations, although not above 1% in the dry matter, were within the range obtained by McGrath et al. (1993) as well as that measured by Brown et al. (1995b) and Meerts & Van Isacker (1997) in Thlaspi populations originating from Prayon and Plombières. They were higher than those measured by Ebbs et al. (1997) in Thlaspi grown in soil with higher Zn and Cd concentrations than the Dornach and Giornico soils. This indicates that our samples have Zn concentrations commonly observed for T. caerulescens in a wide range of soil conditions.
Elemental distribution at the subcellular level
Our micro-analytical data on high-pressure-frozen and cryosectioned leaves of T. caerulescens show for the first time that the cells from the stomatal complex contained distinctly lower zinc concentrations in the vacuoles compared with the vacuoles of epidermal cells. Zinc was mainly localized in elongated epidermal cells in the upper and lower epidermis of the leaves. Preferential accumulation of Zn in the epidermal cells was also observed in the study of Küpper et al. (1999) , with T. caerulescens containing total Zn concentrations of up to 306 mmol kg−1 dry mass in the leaves and in the study of Vázquez et al. (1994) containing total Zn concentrations of up to 92 mmol kg−1 dry mass in the leaves. Küpper et al. (1999) reported that the relative Zn concentration in epidermal cells correlated linearly with the cell length, suggesting that vacuolation of epidermal cells promoted the Zn accumulation within the cells. However, the authors did not distinguish between different cell compartments (vacuoles versus cell walls) and between epidermal cells and cells from the stomatal complex in the freeze-fractures.
Lowering the Zn content in the cells of the stomatal complex indicates that these cells have little Zn tolerance and may have membrane properties distinct from other epidermal cells. Identification of guard cells in the cryosections was facilitated by the high concentrations of K in comparison with the adjacent cells ( Marschner 1995) and the intracellular structures as well as the abundance of starch granules (observed at higher magnification) were similar to the structures of guard cells shown by Vázquez et al. (1994) . Guard cells and subsidiary cells are known to have a special function in stomatal regulation. Stomatal movements are mediated by the accumulation and release of ions (K+, balanced by Cl− and malate from starch) in and from the vacuoles of the guard cells ( Hite & Outlaw 1994). The specific ion-transport properties of cells from the stomatal complex may be able to partly exclude zinc from the stomata, thus making this potentially toxic metal unavailable for interaction with metabolically active cellular compartments. A similar mechanism to protect chloroplasts from the toxicity of Zn (exclusion of Zn in chloroplasts) could occur in the mesophyll cells. Nickel showed a different pattern of distribution in the epidermal cells of leaves of T. montanum var. siskiyouense ( Heath, Southworth & D’Allura 1997). Compartmentation of Ni was found in subsidiary cells that surround guard cells but not in guard cells or in epidermal cells. This difference might be related to the different behaviour of Ni as opposed to Zn, the different Thlaspi species or the different preparation procedure used for EDXMA. Measurements were carried out on chemically fixed leaf surfaces and not in freeze-fractures or in cryosections.
Vázquez et al. (1994) , using EDXMA at room temperatures, localized Zn in electron-dense deposits in the vacuoles of epidermal and subepidermal cells of the leaves of T. caerulescens. In our study, however, Zn stored in the vacuoles of the leaf epidermis was evenly distributed and no Zn-containing crystals or deposits were observed in high-pressure-frozen, cryosectioned plant material. High concentrations of Zn were often accompanied by high concentrations of Ca in the epidermal cells, and Ca was the most abundant element in the vacuoles of mesophyll cells. The role of the relatively high Ca concentrations in vacuoles of epidermal and of mesophyll cells in T. caerulescens is not clear. Accumulation of Ca in leaves seems to be plant-family specific, and the Ca concentration in the soil solution may influence the relative distribution of Ca between mesophyll and epidermal cells ( Dietz et al. 1992 ). The high concentrations of cations in the vacuoles raises the question of which anions counterbalance the high Zn and Ca concentrations inside the vacuoles. Neither Cl, S nor P were present in appreciable quantities, and therefore anions of these elements alone would not have been sufficient to balance the high concentrations of cations. There was also no evidence that Zn was precipitated with inorganic P as suggested as a Zn detoxification mechanism in the root vacuoles of Deschampsia caespitosa ( Van Steveninck et al. 1987 ). Recently, Salt et al. (1999) determined the ligand environment of Zn in different tissues of T. caerulescens using X-ray absorption spectroscopy. They demonstrated that organic acids are involved in xylem transport and Zn storage in shoots and the Zn coordination in the shoots was dominated by citrate with smaller contributions from hydrated zinc, histidine and oxalate binding.
Vacuolar Zn sequestration in epidermal cells, in the present study, was not the only tolerance mechanism in leaves of T. caerulescens. The high levels of Zn binding to epidermal cell walls and intercellular spaces between epidermal and mesophyll cell walls in this study are in good agreement with the findings of Vázquez et al. (1994) , indicating that the accumulation of Zn in the cell wall and intercellular spaces of epidermal cells is an important tolerance mechanism for Zn in T. caerulescens. Whether the Zn located in the intercellular spaces is an artefact or not is unclear from this study. The leaf intercellular spaces would normally be air-filled. Here, however, it was necessary to replace intercellular gases by 1-hexadecene before high-pressure freezing. Gas bubbles in the intercellular spaces would collapse when subjected to high pressure ( Michel et al. 1991 ). Another source of artefacts can be introduced by cryofixation. Jeffree et al. (1987) demonstrated that water may be displaced from cells of Phaseolus vulgaris into intercellular spaces prepared by conventional freezing. However, extracellular deposits on the surfaces of cells have not been observed in high-pressure-frozen material.
High Zn accumulation in the leaves is partly due to the binding of the cationic Zn to the anionic sites in the cell wall. Similarly, Salt et al. (1999) , found a substantial fraction of Zn in the leaves of T. caerulescens to be complexed to the cell wall, suggesting that cell wall binding sites have a high affinity for Zn, thereby lowering potentially toxic-free Zn ions in the cytoplasm. Zinc exudation as metal-rich viscous droplets of leaves was recently proposed as a Zn tolerance mechanism (Leblanc, published in Brooks 1998). Using SEM-EDXMA he observed small hemispherical bodies containing Zn and S at the surface of some leaves of T. caerulescens. In the present study no metal-rich droplets were observed. Both the adaxial and the abaxial leaf surfaces were covered with epicuticular wax crystals only, suggesting that these structures are a constitutive property of the T. caerulescens species used in this experiment.
In the roots, both the cell walls and the vacuoles of epidermal and cortical cells represent sites for the localization of Zn ( Vázquez et al. 1992 , 1994). Salt et al. (1999) found that the majority of the intracellular Zn in roots of T. caerulescens was coordinated with histidine, with the rest complexed to the cell wall. In our study, binding of Zn to negatively charged sites associated with the cell wall contributed substantially to the binding of Zn accumulated in roots. However, the binding property of root cell walls seems unlikely to be significant, since most of the Zn was transported into the shoot (accumulation five times higher than in roots). In contrast to Vázquez et al. (1992) , we did not find that Zn had accumulated in the root vacuoles of T. caerulescens. However, in their study the total Zn concentrations in roots reached 60 mmol kg−1 dry mass, four times higher than in our study. The apparent accumulation of S in the vacuoles of root cortical cells, in the stele and the veins of the leaves raises the question of whether sulphur is involved in Zn tolerance and hyperaccumulation. Phytochelatins are important S-containing chelates known to detoxify higher plants from certain heavy metals. However, a key role of phytochelatins in Zn tolerance mechanism is questionable ( Rauser 1990). From this study, binding of Zn to sulphhydryl compounds can probably be rejected since there was no evidence of association between Zn and S.
From our results we conclude that T. caerulescens has a specific aptitude for taking up Zn. This is expressed by the high general Zn uptake in leaves whatever the level of soil contamination and also the ability to remove forms of Zn more tightly bound to the soils than those extractable with NaN03 (0·1 mol dm−3). Zinc accumulation in leaves was limited to peripheral epidermal cells. Observed patterns of Zn distribution were the same regardless of the inital Zn content of the soil or the total Zn content of the plant. Both the cell walls and the vacuoles were involved in the Zn tolerance mechanisms to protect metabolically active cellular compartments from toxic Zn concentrations. Vacuolar Zn sequestration was predominantly restricted to epidermal cells, whereas guard and subsidiary cells were protected from exposure to deleterious concentrations of this metal. It would be rewarding to study the element distribution over a full growing season in order to determine the re-allocation of elements within the plant and the mechanisms involved in the process. Further elucidation of the specifity of induced Zn-binding processes in functionally different epidermal cells remains a challenge.