High-resolution elemental localization in vacuolate plant cells by nanoscale secondary ion mass spectrometry


*For correspondence (fax +44 1865 283217; e-mail chris.grovenor@materials.ox.ac.uk).


By combining the capabilities of advanced sample preparation methodologies with the latest generation of secondary ion mass spectrometry instrumentation, we show that chemical information on the distribution of even dilute species in biological samples can be obtained with spatial resolutions of better than 100 nm. Here, we show the distribution of nickel and other elements in leaf tissue of the nickel hyperaccumulator plant Alyssum lesbiacum prepared by high-pressure freezing and freeze substitution.


One of the greatest technical challenges in cell biology is to produce an accurate spatially resolved image of the distribution of chemical elements at the subcellular level. A number of chemical analytical techniques have been applied to the investigation of elemental localization in cells. Considerable interest in recent years has centred on studies of trace element distribution in plant tissues, using techniques such as scanning electron microscopy with energy-dispersive X-ray analysis (SEM–EDX) (e.g. Psaras et al., 2000; Küpper et al., 2001; Bidwell et al., 2004; Broadhurst et al., 2004a,b), X-ray photoelectron spectroscopy (XPS), electron energy loss spectroscopy (EELS) (Lichtenberger and Neumann, 1997), proton-induced X-ray emission (PIXE) (Mesjasz-Przybyłowicz et al., 1994; Krämer et al., 1997) and synchrotron X-ray fluorescence (SXRF), combined with computer microtomography (McNear et al., 2005a,b; Punshon et al., 2005, 2009; Lombi and Susini, 2009). These techniques, however, are all limited in their applications in biology by an intrinsic compromise in the achievable spatial resolution of dilute elements. Whereas thin sections provide the highest possible spatial resolution, the intensity of chemical signals from such samples will often be below the detection limit of the instrumentation; conversely, larger signals can be obtained from thicker specimens, but with an associated loss of spatial resolution from beam spreading effects. As a consequence, these techniques have been limited to a spatial resolution of about 1 μm. Although this has proved useful in mapping elemental localization at the tissue and cellular levels, new approaches capable of achieving higher spatial resolution are needed if the distribution of elements is to be revealed at the subcellular level using chemical mapping techniques.

Over 40 years ago, secondary ion mass spectrometry (SIMS) was introduced as a surface analysis technique in which a focused primary ion beam is scanned across a sample (Castaing and Slodzian, 1962; Liebl, 1967). The bombardment of these primary ions results in the ejection of charged atomic and molecular species from the surface layers of the sample. These secondary ions are then separated on the basis of their mass-to-charge ratio using a high-performance mass spectrometer, and are correlated with their spatial origin to form a chemical image. SIMS was first applied to biological material more than 30 years ago (Galle, 1970), and the technique has subsequently been applied to plants in a number of studies (Grignon et al., 1997; Dérue et al., 2002, 2006a,b; Heard et al., 2002; Laurie et al., 2002; Mills et al., 2005). Grignon et al. (1996) particularly commented on the important role of SIMS analysis in trying to resolve the distribution of key elements in the cell walls, cytoplasm and central vacuole, for which submicrometer resolution is required. However, data from the instruments available in these previous studies were also limited by compromises between sensitivity and resolution.

The NanoSIMS is the latest generation of dynamic (high-intensity) SIMS instrumentation, and is capable of achieving high spatial resolution (<50 nm) with excellent sensitivity, and, at the same time, the high mass resolving power (m/Δm approximately 5000) needed to discriminate clearly between the isobaric ionic species commonly produced from the bombardment of biological materials (Slodzian et al., 1992; Hillion et al., 1993). The combination of state-of-the-art ion optics and very high-efficiency detectors results in a parts per million sensitivity or better for many elements. Many recent applications of the NanoSIMS instruments have been in the fields of geology (Stern et al., 2005; Fletcher et al., 2008; Orphan and House, 2009), cosmochemistry (Hoppe, 2006), materials science (Kailas et al., 2006) and soil science (Hermann et al., 2007). Recent uses of NanoSIMS in biology have included experiments on lipid membranes (Kraft et al., 2006), on the direct correlation of fluorescence imaging and NanoSIMS chemical analysis (Behrens et al., 2008), on the distribution of Fe-containing antigens (Dauphas et al., 2008) and on the development of stable-isotope analysis with high spatial resolution (Lechene et al., 2006) which has enabled novel studies of microbial metabolism (Lechene et al., 2007; Musat et al., 2008; Finzi-Hart et al., 2009; Wagner, 2009). Moore et al. (2010) recently reported on the use of NanoSIMS to map the distribution of arsenic and selenium in dried cereal grains which, on account of their low water content, required little in the way of specimen preparation.

Grignon (2007) has recently reviewed the use of state-of-the-art SIMS instruments in the chemical analysis of higher plants. The ultra-high vacuum requirement of SIMS experimentation, which is needed to prevent the scattering of the primary and secondary ion beams, and to stop adsorption of impurities onto the sample surface (Vickerman, 1989), means that samples must be fixed as close as possible to their in vivo state. This constraint represents a major challenge in the preparation of biological materials for microanalysis, as has been extensively discussed in the literature (e.g. Burns, 1982; Chandra and Morrison, 1992; Boxer et al., 2009). We note that quantification of chemical analysis in SIMS analysis can also be problematic because the secondary ion signal generated depends not only on the concentration of the element of interest in the sample, but also on the ionization probability of the element, and the nature of the matrix in which the element is embedded (Benninghoven et al., 1987).

It is now generally accepted that ultra-rapid cryofixation is the preferred strategy for stabilizing biological material close to the in vivo state. Various techniques have been developed for freezing biological material and preventing the formation of ice crystals in the cytoplasm. For plant tissue samples with high water content, high-pressure freezing is the most suitable technique, and can be combined with freeze-substitution protocols for thin sectioning of resin-embedded material (Kiss et al., 1990; Hess, 2007). This technique was thus chosen as the most appropriate for the preparation of hydrated biological material for NanoSIMS analysis.

In the present study, the nickel hyperaccumulator plant Alyssum lesbiacum was used to evaluate the potential for revealing elemental localization in highly vacuolated plant material at the subcellular level using NanoSIMS. This plant species has been used previously in studying elemental localization at the tissue and cellular levels, showing, for example, that nickel accumulates to its highest concentrations in non-photosynthetic epidermal cells in the shoot (e.g. Krämer et al., 1997; Küpper et al., 2001), and that Ni and Ca can reach concentrations as high as 1 and 8% (w/w), respectively, in the epidermal trichomes (Smart et al., 2007). In principle, the NanoSIMS technique has the high chemical sensitivity and spatial resolving power needed to study the sequestration of elements at the subcellular level in highly vacuolate, metabolically active cells in the leaf, but elemental concentrations can be much lower in such cells, and the preservation of detailed cell ultrastructure during sample preparation is more demanding. As well as the 58Ni signal, we also selected the 16O, 12C14N, 31P and 32S signals to assist in interpreting the distribution of elements between different subcellular compartments. The CN ion can be used in the SIMS analysis of biological materials as a marker for the distribution of proteins and nucleic acids (Levi-Setti, 1988), and to highlight the morphology of the sample (Grignon et al., 1992), whereas O ions are generated preferentially from structures such as cell walls rich in cellulose. P ions show the distribution of molecules such as nucleic acids and phospholipids, and S ions map the S-containing proteins (Guerquin-Kern et al., 2005; Quintana et al., 2006). In combination, these signals should have the potential to provide information on the ultrastructural localization of important elements in biological samples at high spatial resolution.


A preparative procedure involving high-pressure freezing (HPF), followed by freeze substitution (FS) with acetone, and resin embedding was used to preserve the ultrastructure of leaves of the nickel hyperaccumulator plant A. lesbiacum, and maintain as faithfully as possible the in vivo spatial distribution of elements during sample processing. Serial sections of HPF/FS-embedded A. lesbiacum material were analysed by optical microscopy, NanoSIMS and transmission electron microscopy (TEM). Reflectance optical microscopy of 0.5-μm-thick cross sections of the linear leaves mounted on the silicon NanoSIMS sample holder show turgid cells with regular outlines (Figure 1a). Highly vacuolate epidermal cells and chloroplast-containing mesophyll cells are clearly identifiable, as are the cross sections of two vascular bundles and parts of several trichomes. Although there is some evidence of mechanical damage (e.g. where one ray of a stellate trichome has fractured near its centre), overall the level of preservation is extremely high (cf. electron micrographs in Figures 2e,f and 3e,f). A section cut further into the leaf tissue shows a unicellular trichome sectioned through its radial longitudinal axis, revealing the stalk and swollen basal region of the trichome embedded in the epidermal cell layer (Figure 1b).

Figure 1.

 Reflectance optical micrographs of a resin-embedded cross section of nickel-rich Alyssum lesbiacum leaf prepared by high-pressure freezing, followed by freeze substitution, showing a high level of tissue preservation.
(a) Part of a transverse leaf cross section showing trichomes (T), epidermis (E), mesophyll (M) and vascular bundles (V). Scale bar: 100 μm.
(b) Section through the radial longitudinal axis of an epidermal trichome showing the trichome base embedded in the epidermal cell layer, and showing the exact area subsequently analysed by NanoSIMS (see Figure 3a). Scale bar: 50 μm.

Figure 2.

 NanoSIMS analysis of a cross-section of a nickel-rich Alyssum lesbiacum leaf prepared by high-pressure freezing followed by freeze substitution.
(a) NanoSIMS maps from a peripheral region of the leaf cross section, including a stomatal complex obtained using the Cs+ primary ion beam, and showing 16O, inline image, 12C14N, 31P and 58Ni ion maps, and a colour overlay for Ni and P. Scale bar: 10 μm.
(b) NanoSIMS maps obtained using the O primary ion beam for the distribution of 23Na+, 24Mg+, 39K+, 40Ca+ and 58Ni+ signals, from a region of the leaf surface including a stomatal complex. The secondary electron (SE) image acquired with the Cs beam is included to show the morphology of the imaged region. Scale bar: 10 μm. A heat scale for the images in panels (a) and (b) is shown on the right.
(c) NanoSIMS composition profiles across a single epidermal cell for the inline image, 31P, 12C14N and 58Ni maps shown in (a). The exact position of the line analysed is indicated with an arrow in the SE image in (b). E, epidermal cell; EC, epidermal cytoplasm; EV, epidermal vacuole; EW, epidermal wall; R, resin.
(d) NanoSIMS composition profiles across a single epidermal cell for the 23Na+, 24Mg+, 39K+, 40Ca+ and 58Ni+ maps shown in (b). The exact position of the line analysed is indicated with an arrow in the secondary electron image in (b). E, epidermal cell; EC, epidermal cytoplasm; EV, epidermal vacuole; EW, epidermal wall; R, resin. EW and EC are shown combined because of the lower spatial resolution achieved with the O ion beam.
(e) Transmission electron microscopy (TEM) image showing detailed ultrastructural preservation in a stomatal guard cell. Ch, chloroplast; CW, cell wall; Vac, vacuole. Scale bar: 1 μm.
(f) TEM image showing detail of a vacuole, cortical cytoplasm and cell wall in an epidermal cell. CW, cell wall; Vac, vacuole. Scale bar: 0.2 μm.

Figure 3.

 NanoSIMS analysis of the peripheral region of a leaf cross section of Alyssum lesbiacum prepared by high-pressure freezing followed by freeze substitution, containing the basal portion of a trichome cell embedded in the epidermal cell layer and adjacent to the underlying mesophyll.
(a) NanoSIMS maps obtained using the Cs+ primary ion beam, showing inline image, 12C14N, 31P, 32S and 58Ni ion maps, and a colour overlay for Ni and P. Scale bar: 10 μm. Note the distribution of nickel in the trichome cell wall. Scale bar: 8 μm. A heat scale for these images is shown on the right.
(b) NanoSIMS secondary electron (SE) map obtained using the Cs+ primary ion beam. Ch, chloroplasts; E, epidermal cell; N, nucleus; NO, nucleolus. Scale bar: 8 μm.
(c) NanoSIMS composition profiles across the trichome base for the inline image, 31P, 12C14N, 32S and 58Ni maps shown in (a). The exact position of the line analysed is indicated with an arrow in the secondary electron image in (b). M, mesophyll cell; N, nucleus; NO, nucleolus; R, resin; TB, trichome base; TJ, tripartite junction.
(d) Transmission electron microscopy (TEM) image showing detailed ultrastructural preservation of the tripartite junction between trichome and mesophyll cells. Note the thick lamellate structure of the walls. C, cell wall; V, vacuole. Scale bar: 1 μm.
(e) TEM image showing detailed ultrastructural preservation of the interface between the cell wall and the waxy cuticle in the trichome stalk. Note the platelets embedded in the wall matrix (arrows). C, cell wall; Vac, vacuole. Scale bar: 0.5 μm.

The black and white NanoSIMS secondary electron map (Figure 2b) shows a cross section of a leaf with a stomatal complex (sectioned asymmetrically), epidermal cells and palisade mesophyll cells. The image resolution is sufficiently high to reveal starch grains within a single chloroplast in one of the two stomatal guard cells (cf. Figure 2e), and prominent starch-containing chloroplasts in the mesophyll cells. The high level of structural preservation in these cells is also evident in the electron micrographs from the HPF/FS leaf material (Figure 2e,f). The other chemical maps in Figure 2a,b allow comparison of the local elemental composition between the immediately adjacent stomatal complex, epidermal and mesophyll cells. Using the primary Cs+ ion beam (Figure 2a), the O ion signal is particularly strong from the cell walls in all cell types, whereas the inline image ion signal is relatively homogeneous across different cells, but is especially low from the starch grains, and is elevated from the resin. The CN ion signal is high specifically in the cytoplasm, notably in the mesophyll cells and guard cells, but also in the thin peripheral layer of cytoplasm discernible in the epidermal cells; similarly, the P ion signal clearly marks the metabolically active parts of the cell cytoplasm. Both the subsidiary cells and epidermal cells contain a high concentration and uniform distribution of Ni in their large central cell vacuoles, in contrast to the considerably lower Ni concentration in the guard cells and mesophyll cells. Concentrations of Ni are also elevated in the cell walls of all cell types, which contrasts with the low Ni counts in the thin cortical layer of cytoplasm close to the cell periphery of the epidermal and subsidiary cells (Figure 2f), as well as inside the metabolically active guard cells and mesophyll cells. Indeed, the distribution of Ni and P show a striking inverse relationship (as emphasized by the colour overlay in Figure 2a), with high Ni concentrations in the vacuolar lumen and cell walls contrasting with the P-rich, metabolically active cytoplasm. This relationship was confirmed by line scans across epidermal cells (Figure 2c), in which it is possible to clearly distinguish the cell wall (high CN and moderately high Ni levels), cytoplasm (high CN and very high P levels) and vacuole (low CN, low P and high Ni levels). Applying the conventional definition for edge response in images of this kind (separation between the 10 and 90% intensity contours), the NanoSIMS images in Figure 2 show a spatial resolution in the P and Ni ion signals of better than 80 nm.

Using the primary O ion beam, we were able to acquire maps to show the distribution of 23Na+, 40Ca+, 39K+, 24Mg+ and 58Ni+ ions from the same area analysed with the Cs+ beam (Figure 2b). As expected, the 58Ni+ signal distribution is similar to that observed for 58Ni (Figure 2a), with highest intensities in the vacuole. However, because of the intrinsically poorer resolution of the O beam, the cell cytoplasm cannot be discerned in these images. The high binding capacity of the cell wall matrix for divalent cations such as calcium and nickel is reflected in the epidermal cell line profile (Figure 2d), although it is notable that the calcium signal is much higher in the cell wall than the vacuole, whereas the converse is true for nickel. The other cations Mg, K and Na also show enrichment in the cytoplasmic phases of the cell compared with the vacuole (Figure 2b,d).

At the end of their development, the trichome cells of A. lesbiacum die, but remain on the shoot surface as a protective barrier. A NanoSIMS secondary electron map was made of the basal region of a living trichome sectioned in a radial longitudinal plane (Figure 3b). A large nucleus and several small vacuoles are clearly visible. Trichome cells in the family Brassicaceae are polyploid at maturity, having proceeded through one or more rounds of endoreduplication of the entire genome, so the nucleus in these cells is typically large with a very prominent nucleolus (Melaragno et al., 1993; Hülskamp et al., 1994). Electron micrographs of a trichome base (Figure 3d) and lateral basal wall (Figure 3e) again demonstrate the high quality of the preservation of the material subjected to analysis. The inline image ion signal from this same area is relatively homogeneous, except for the starch grains that, as in Figure 2a, give a very low signal intensity (Figure 3a). As observed previously, the CN ion signal is strong in the protein-rich cytoplasmic parts of the mesophyll cells surrounding the starch grains; in addition, the nucleus and nucleolus also have a strongly elevated CN signal intensity. In all of the inline image, CN and S ion maps, the organelles and cytoplasmic compartments are discernible, and a spatial resolution better than 70 nm is achieved in the CN image. The S signal is elevated in the tripartite wall junctions between the trichome and neighbouring mesophyll cells, in the epidermal cell vacuoles and in the starch grains. The P distribution is often used as an indicator of regions of metabolic activity, and nuclei from both the trichome and an epidermal cell are clearly visible. Furthermore, the nucleolus of the trichome cell has an elevated P signal, presumably reflecting the high investment of P in RNA (Elser et al., 2000; Ågren, 2004); other discrete points of very high P signal may correspond to cross sections through chromatin.

The Ni signal is very low in the interior of the metabolically active trichome cell, but is strong in the tripartite junctions at the base of the trichome, reflecting nickel binding to the cell wall and middle lamella. Transmission electron microscopy reveals the polylamellate nature of these walls (Figure 3d). After being delivered from the root to the shoot in the xylem, nickel is probably swept through the apoplast of the leaf in the transpiration stream, and may accumulate in the middle lamella by binding to the fixed negative charges of galacturonic acid residues in pectin. In addition, there is an elevated, evenly distributed Ni signal from within the epidermal cell vacuoles either side of the trichome ‘stalk’, which has a complex wall structure comprising large platelets of wax embedded in the wall matrix (Figure 3e). The complementary distribution of Ni and P in these cells is once again evident from the merged image in Figure 3a. Line scans across the basal region of the trichome show elevated Ni and S signals in the tripartite junction (TJ), with a clearly elevated P signal in the nucleus (N) and nucleolus (NO) (Figure 3c).


Introduction of the latest generation of high-resolution imaging by secondary ion mass spectrometry has opened up the potential to map the spatial distribution of elements at the submicrometer level. However, it has become clear that realizing the full potential of this technique in biology depends critically on the ability of sample preparation techniques to preserve as faithfully as possible the original distribution of elements in situ while removing water from the tissue. In the present study, the combination of rapid HPF of the tissue samples (210 MPa at −196°C for a duration of 30 ms) followed by FS in acetone was found to produce excellent structural preservation of the leaf tissue of A. lesbiacum. This is demonstrated, for example, by the homogeneous distribution of nickel observed in the vacuole, notably in epidermal cells, which is consistent with the chelation of nickel in the soluble phase by ligands such as carboxylic acids (Krämer et al., 2000; Saito et al., 2010).

Another feature that demonstrates the high resolving power of NanoSIMS when combined with appropriate sample preparation is the resolution of elemental concentrations in different subcellular compartments, particularly in images obtained with the Cs+ primary ion beam. This is particularly well exemplified by the contrast between high Ni in the vacuolar and cell wall phases of the epidermal cells of A. lesbiacum and low Ni in the cell cytoplasm (Figure 2a), consistent with the prediction that free Ni2+ concentrations should be in the picomolar range in cytoplasmic phases (Williams, 2007). Intracellular Ni concentrations are evidently much lower in the stomatal guard cells and most of the palisade mesophyll cells, which are metabolically highly active and less vacuolate. A number of other ions such as 12C14N, inline image and 31P provided additional information about protein, lipid and nucleic acid distribution at the subcellular level. These results refute the view that NanoSIMS imaging may not be able to provide information on the compartmentation of elements at the subcellular level (Slaveykova et al., 2009).

Although considerable research has been performed on nickel hyperaccumulator plants at the cellular level, there is currently no consensus on the principal sites of nickel localization. Subcellular fractionation reports suggest 73 and 20% of the nickel in the leaves of Thlaspi goesingense is associated with the cell wall and vacuole, respectively (Krämer et al., 2000), but reports from cryo-SEM–EDX indicate that nickel is localized primarily in vacuoles of the epidermal cells rather than the cell walls of Alyssum bertolonii, A. lesbiacum and T. goesingense (Küpper et al., 2001). In SEM–EDX of dried herbarium samples, nickel was apparently excluded from the guard cells and subsidiary cells, but was present in higher concentration in epidermal cells (Psaras et al., 2000). In a high-resolution SEM–EDX study of freeze-substituted Hybanthus floribundus, the epidermal vacuole was found to be an important site of nickel accumulation, but nickel was also found to be evenly distributed in the outer cell wall phase of epidermal and mesophyll cell tissue throughout the leaf (Bidwell et al., 2004).

Plants in the genus Alyssum display characteristic large unicellular stellate trichomes. Cryo-SEM–EDX analysis indicated the exclusion of nickel from the bulk of the trichomes of A. bertolonii and A. lesbiacum, although subsequent staining with dimethylgloxime suggested a locally high concentration of nickel in the trichome base (Küpper et al., 2000). In contrast, PIXE of frozen dehydrated and fixed A. lesbiacum indicated nickel is sequestered to a considerable degree in the upper portion of the trichome (Krämer et al., 1997). In SEM–EDX of dried herbarium samples, nickel was apparently excluded from the trichomes (Psaras et al., 2000). In a further SEM–EDX study of Alyssum murale prepared by freeze drying, nickel was reported to be stored only in the basal portion of the trichome (Broadhurst et al., 2004a). SXRF and computed microtomography (CMT) of shock-frozen and partially dried leaves of A. murale has been performed, and the authors suggest nickel is concentrated in the basal portion of the trichome; however, with a spatial resolution of 50 μm, limited by the absorption edge CMT obtained using living plants, it was difficult to discern individual cells (McNear et al., 2005a,b). Our previous work (Smart et al., 2007) with an electron probe microanalyser (EPMA) and preliminary NanoSIMS analysis suggests strong localization of Ni to the trichome arms, but we were unable to make confident statements about Ni concentrations in the epidermal cell vacuoles or trichome basal portions.

The NanoSIMS data obtained from the HPF–FS material represents a major advance in terms of high-resolution chemical analysis. We show clear evidence for significant sequestration of nickel in the epidermal cell vacuoles, the upper portion of the trichome and in basal tripartite junctions, with a uniform distribution throughout the cell walls of all the cells in the sections observed. The proposed tolerance mechanisms of hyperaccumulating plants involved active sequestration in the vacuole and cell wall (Pilon-Smits, 2005). Nickel was found to be excluded from the protoplast of stomatal cells, trichome bases, vascular bundles and mesophyll cells. This pattern of sequestration presumably serves to protect the enzymes of central cell metabolism contained in the cytoplasm, which can be inactivated by metal ions (indeed the phosphorus map is the inverse of the nickel map). The distribution of nickel in the vacuole is uniform, suggesting that it is associated with water-soluble compounds such as carboxylic acids (Bidwell et al., 2004; Pilon-Smits, 2005).

Elevated concentrations of calcium have been reported in the trichomes of A. lesbiacum (Krämer et al., 1997; Psaras et al., 2000; Küpper et al., 2001; Broadhurst et al., 2004a,b; Smart et al., 2007), and the data also indicated that calcium was strongly localized in the upper portion of the trichome. High concentrations were also observed here by NanoSIMS in the cell wall, particularly of epidermal (Figure 2b) and trichome cells (data not shown).

Although the NanoSIMS has sufficient spatial resolution and analytical sensitivity to allow the distribution of elements in these HPF–FS specimens to be mapped at the subcellular level, the technique does have some restrictions. It is very sensitive to topography, and so flat, microtomed specimens are required. SIMS is subject to strong matrix effects, and hence signals from a matrix as heterogeneous as plant cells will be affected. For instance, the yield of carbon ions is affected in very different ways from a semicrystalline carbohydrate starch grains and the mainly resin matrix of the vacuole, although the carbon content in these two different regions is likely to be rather similar. In spite of this, the very different local yields of the selected ions in the data presented suggests strongly that the composition maps represent true variations in chemistry, and are not a result of simple differential yield or topographical effects. This is especially apparent in the P images where, as expected, high levels are found in the cytoplasm, and S images where the localization to protein-containing cell walls is very clear. Furthermore, the TEM analysis established the correct interpretation of histological features in the NanoSIMS chemical maps, and also demonstrated that the structural preservation of the material was extremely high.

Another concern with biological material is the method used to preserve the specimens prior to analysis. The HPF–FS technique applied here (ultra-rapid freezing and slow substitution of ice with chemical fixatives) helps to minimize the redistribution of soluble elements, but the extent of possible elemental redistribution during sample preparation is an important factor in the preparation of biological specimens for any kind of chemical analysis. Analysis of the data on the distribution of electropositive elements in Figure 2b shows that the in vivo distributions of highly diffusible ions such as 23Na+, 24Mg+, 39K+ and 40Ca+ have been significantly altered during chemical fixation. For instance the K/Ca ratio in the epidermal cell vacuoles is much lower than we would expect from a living cell (Dérue et al., 2006b). It is, however, more likely that the distribution of ions like inline image, 12C14N and 31P will remain relatively unmodified by resin impregnation if they are the characteristic ionic fragments from the interaction of the incident analysis beam with large immobile molecules such as lipids, proteins and DNA (Heard et al., 2002). However, this does not explain the apparent preservation of nickel in its in vivo state. The nickel–organic acid complex should be highly mobile, so the preservation in situ testifies to the effectiveness of the HPF–FS protocol in preserving the distribution of soluble elements. Part of the nickel detected in the cell wall may be physically bound to negatively charged sites on the cell wall (and may displace some of the extracellular bound calcium), and thereby be rendered less mobile (Krämer et al., 2000).

Experimental procedures

Plant cultivation

Plants of the nickel hyperaccumulator A. lesbiacum (Candargy) Rech.f. (Brassicaceae) were grown from seed collected from an ultramafic (serpentine) soil on Lesbos, Greece. After germination on moist sand, plants were grown hydroponically for 4 months in a controlled-environment glasshouse in modified 0.1-strength Hoagland solution (Ingle et al., 2005) supplemented with 300 μm NiSO4. Mature leaves from these plants were analysed by atomic absorption spectrophotometry following wet acid digestion in 69% (v/v) HNO3 according to the procedure of Roosens et al. (2003), and were determined to have a bulk Ni content of 7.87 ± 1.89 g per kg dry shoot biomass (mean ± SD, n = 4 leaves).

Sample preparation

Healthy leaves of A. lesbiacum were selected, and 2-mm-long segments were dissected from each leaf using a razorblade and then placed in a 3-mm-diameter copper planchette. Prior to capping the planchette, the sample was covered with the cryoprotectant hexadecene. Pairs of planchettes were clamped together and immediately frozen using a Bal-Tec HPM 010 high-pressure freezer (Bal-Tec AG, now part of Leica Microsystems, http://www.leica-microsystems.com), which subjected the sample to a pressure of 210 MPa at −196°C for a duration of 30 ms.

The FS with acetone was carried out in a Reichert AFS (Leica) FS system. The procedure used for FS was derived from a combination of the protocols developed by Steinbrecht and Muller (1987) for the FS of insect material, and by Studer et al. (2001) for the FS of plant and animal tissue. The planchettes were split open under liquid nitrogen and placed in microporous specimen pots containing the acetone previously frozen in liquid nitrogen. Specimen pots were placed in a universal aluminium container on the surface of the frozen acetone substitution medium, and were then transferred into the FS unit pre-cooled to −160°C.

The FS system was programmed to 30 min at −160°C, followed by heating at a rate of 15°C h−1 to −85°C, and then 24 h at −85°C. At this stage the acetone slush was replaced with an acetone and 2% (w/v) osmium tetroxide mixture to fix the sample. The FS machine was then programmed as follows: 26 h at −85°C, heating rate of 2°C h−1 to −60°C, 8 h at −60°C, heating rate of 2°C h−1 to −30°C, 24 h at −30°C, heating rate of 1°C h−1 to −20°C, 24 h at −20°C. On completion, samples were removed from the FS machine and placed at 4°C for 24 h.

All subsequent steps were carried out at room temperature (20°C). Samples were rinsed in anhydrous acetone three times, each time for 20 min. Samples were then carefully removed from the microporous pots and embedded gradually in Spurr resin without accelerator. A graded resin/acetone (v/v) series was used: 10, 25, 50, 75 and 100% resin, with each step lasting 2 h. Samples were then placed in 100% resin without accelerator for 8 h and in 100% resin with accelerator for 8 h; this step was repeated five times. Samples were then placed in moulds containing fresh resin and were polymerized for 9 h at 70°C.

Sections of 0.5 μm thickness were cut using a diamond knife on an RMC Powertome Ultramicrotome (Boeckeler Instruments, http://www.boeckeler.com) and mounted onto 7 × 7 mm silicon squares for NanoSIMS analysis. An Axioskop 2 MAT light microscope equipped with an AxioCam MRc5 digital camera (Carl Zeiss MicroImaging GmbH, http://www.zeiss.de/micro) was used in reflectance mode to assess the quality of structural preservation in samples prior to gold coating and subsequent NanoSIMS analysis. An SCD 040 sputter coater (Balzers, now part of Oerlikon, http://www.oerlikon.com) was used to coat samples prepared for NanoSIMS. A time of 30 sec with a cathode–specimen distance of 30 mm and deposition current of 30 mA were used to produce a coating thickness of 30–50 nm.

Ultrathin sections of 90 nm thickness were also cut and floated onto copper grids coated with formvar (Agar Scientific, http://www.agarscientific.com) for TEM analysis. Samples were stained in 1% (w/v) uranyl acetate in methanol for 15 min, washed in methanol and stained with lead citrate. Micrographs were obtained with a JEOL 1200EXII TEM at 120 kV using a Gatan (Gatan, http://www.gatan.com) Dual Vision CCD camera.


A CAMECA NanoSIMS 50 (CAMECA, http://www.cameca.com) was used for high-resolution chemical mapping. The instrument was calibrated using standard calibration samples of silicon patterned with square mesas, for use with the Cs+ primary ion beam, and aluminium covered with a copper grid, for use with the O primary ion beam. The standards were used to calibrate the five detectors for selection of ions of known mass and relative yield. For instance, the silicon standard sample was used to tune detectors 1–5 for the following masses: 16O, 28Si, 30Si, inline image and inline image. In addition, the square pattern of the mesas was used to focus and correct any astigmatism. The trolleys were then carefully calibrated for masses of interest using a bulk nickel metal standard.

High-resolution mass spectra around mass 58 comparing both (i) a control plant grown with no deliberate exposure to Ni and (ii) the nickel-rich plant with the nickel metal standard were taken in the Cs+ primary ion beam mode. The 58Ni peak from the nickel standard aligned precisely with the first of several overlapping peaks from the plant sample. Taking into account the significant mass deficit for nickel (approximately Δm = −80 mmu), the detector tuning for 58Ni can be positioned with confidence. The 58Ni peak from the nickel standard does not coincide with the first of the two overlapping peaks from the nickel-free sample.

In order to obtain high-resolution chemical images of the leaf sections, selected areas of approximately 50 × 50 μm were first rastered for 2 h with a Cs+ primary beam with a current of 30 nA to implant Cs into the sample surface. This is necessary to remove surface contamination, and to reduce the work function of the surface to encourage a high yield of secondary ions. Then the primary aperture (D1 = 3) was introduced to reduce the beam current to 300 pA and the incident probe diameter to about 50 nm, and images of the five selected ions were acquired over an area of 40 × 40 μm, at a resolution of 512 × 512 pixels. This image capture took approximately 1 h. The O primary beam was then used to image the same regions of interest to capture the positive secondary ions. An O primary beam current of 125 nA was used for analysis with D1 = 3. Because of the reactivity of the O ions, no prior implantation is required to encourage a high secondary ion yield. The final images were acquired at a resolution of only 256 × 256 pixels because the oxygen incident beam can only be focused to a spot size of about 200 nm.


The NanoSIMS was funded by an EPSRC grant (GR/M61023), and support during this project was provided under grant number GR/T19797/01 to CG. CH acknowledges the BBSRC for support in HPF and TEM (grant no. BB/D001080/1).