A proteomics approach to investigate the ... - Wiley Online Library

The Plant Journal (2013) 73, 131–142

doi: 10.1111/tpj.12022

A proteomics approach to investigate the process of Zn hyperaccumulation in Noccaea caerulescens (J & C. Presl) F.K. Meyer Thomas Schneider1,*, Daniel Pergament Persson2, Søren Husted2, Maja Schellenberg1, Peter Gehrig3, Youngsook Lee4, Enrico Martinoia1,4, Jan K. Schjoerring2,* and Stefan Meyer1 1 Institute of Plant Biology, Department of Molecular Plant Physiology, University of Zurich, Zollikerstraße 107, 8008, Zurich, Switzerland, 2 Department of Plant and Environmental Sciences, Faculty of Science, University of Copenhagen, Thorvaldsensvej 40, DK1871 Frederiksberg C, Copenhagen, Denmark, 3 Functional Genomics Center, University and ETH Zurich, Winterthurerstraße 190, 8057, Zurich, Switzerland, and 4 Postech-UZH Global Research Laboratory, Division of Molecular Life Sciences, Pohang University of Science and Technology, Pohang, Korea Received 9 February 2012; revised 6 September 2012; accepted 10 September 2012; published online 26 October 2012. *For correspondence (email: [email protected]; [email protected]).

SUMMARY Zinc (Zn) is an essential trace element in all living organisms, but is toxic in excess. Several plant species are able to accumulate Zn at extraordinarily high concentrations in the leaf epidermis without showing any toxicity symptoms. However, the molecular mechanisms of this phenomenon are still poorly understood. A state-of-the-art quantitative 2D liquid chromatography/tandem mass spectrometry (2D-LC-MS/MS) proteomics approach was used to investigate the abundance of proteins involved in Zn hyperaccumulation in leaf epidermal and mesophyll tissues of Noccaea caerulescens. Furthermore, the Zn speciation in planta was analyzed by a size-exclusion chromatography/inductively coupled plasma mass spectrometer (SEC-ICPMS) method, in order to identify the Zn-binding ligands and mechanisms responsible for Zn hyperaccumulation. Epidermal cells have an increased capability to cope with the oxidative stress that results from excess Zn, as indicated by a higher abundance of glutathione S-transferase proteins. A Zn importer of the ZIP family was more abundant in the epidermal tissue than in the mesophyll tissue, but the vacuolar Zn transporter MTP1 was equally distributed. Almost all of the Zn located in the mesophyll was stored as Zn– nicotianamine complexes. In contrast, a much lower proportion of the Zn was found as Zn–nicotianamine complexes in the epidermis. However, these cells have higher concentrations of malate and citrate, and these organic acids are probably responsible for complexation of most epidermal Zn. Here we provide evidence for a cell type-specific adaptation to excess Zn conditions and an increased ability to transport Zn into the epidermal vacuoles. Keywords: Zn hyperaccumulation, Noccaea caerulescens, Zn speciation, proteomics, nicotianamine, zinc.

INTRODUCTION Zinc (Zn) is an essential trace element in all organisms, and is responsible for structural and functional aspects of proteins at all developmental and metabolic stages (Vallee and Auld, 1990). It acts as a co-factor for a large number of proteins and enzymes (Andreini et al., 2006; Broadley et al., 2007). However, excess Zn causes severe problems in plant development, such as chlorosis and root growth inhibition. It has been proposed that these effects are © 2012 The Authors The Plant Journal © 2012 Blackwell Publishing Ltd

caused by replacement of other essential metals such as iron (Fe) and magnesium (Mg) that can be easily substituted by Zn (Marschner, 1995). Also, Zn excess may interfere with a range of vital processes, such as DNA replication, enzyme activities and protein folding and function (Broadley et al., 2007). To protect plants from these negative effects, intracellular Zn concentrations must be tightly regulated. This is achieved by a set of ion transporters and 131

132 Thomas Schneider et al. mechanisms that chelate excess ions in order to store, redistribute and detoxify them (Verbruggen et al., 2009; Kra¨mer, 2010). In a few taxa, the ability to accumulate metals in the above-ground biomass has evolved. This interesting phenomenon is known as metal hyperaccumulation. Most hyperaccumulators are members of the Brassicaceae family (Kra¨mer, 2010). Two species of this family have been shown to be excellent model systems for Zn hyperaccumulation: Noccaea caerulescens (formerly Thlaspi caerulescens) and Arabidopsis halleri (Milner and Kochian, 2008; Verbruggen et al., 2009; Kra¨mer, 2010). Zn hyperaccumulation and tolerance is constitutive to a certain extent at the species level (Becher et al., 2004). General traits in the process are: (i) strongly enhanced loading of the metal ion to the xylem of the root tissue, resulting in increased root-to-shoot transport along the transpiration stream (Lasat et al., 1996, 1998; Hanikenne et al., 2008), (ii) highly efficient metal detoxification and sequestration, mainly in the vacuoles of leaf cells (Ku¨pper et al., 1999; Kra¨mer et al., 2000). Some of these properties are due to constitutive up-regulation of components of the Zn homeostasis machinery that is also present in non-hyperaccumulating species (Clemens, 2006; van de Mortel et al., 2006; Talke et al., 2006; Kra¨mer et al., 2007). At least in some cases, this up-regulation is mostly achieved by the presence of multiple gene copies in the hyperaccumulating plant genome (Dra¨ger et al., 2004; Hanikenne et al., 2008; Verbruggen et al., 2009; O’Lochlainn et al., 2011; Ueno et al., 2011). Several transporters from the zinc/iron-regulated transporter protein (ZIP), P-type ATPase, cation diffusion facilitator (CDF) and plant cadmium resistance (PCR) transporter families are involved in Zn transport (Pence et al., 2000; Desbrosses-Fonrouge et al., 2005; Hanikenne et al., 2008; Song et al., 2010). With respect to Zn hyperaccumulation in leaves, the Zn+/H+ antiporter metal transporter 1 (MTP1) is of special interest because it transports excess Zn from the cytosol into the vacuole in the shoot tissue (Kobae et al., 2004; Desbrosses-Fonrouge et al., 2005; Gustin et al., 2009; Ku¨pper and Kochian, 2010). MTP1 expression in the Zn-hyperaccumulator A. halleri is constitutively higher than in the non-hyperaccumulator Arabidopsis lyrata due to a higher gene copy number (Dra¨ger et al., 2004). In addition, transcript levels of nicotianamine synthase (NAS), which is responsible for synthesis of the putative Zn-binding ligand nicotianamine (NA), are constitutively higher in A. halleri than the non-accumulator Arabidopsis thaliana (Weber et al., 2004; Talke et al., 2006). In addition, Zn-binding ligands such as citrate and malate have also been suggested to contribute to hyperaccumulation (Salt et al., 1999; Sarret et al., 2002; Callahan et al., 2006). In order to study Zn hyperaccumulation, we chose Noccaea caerulescens (Nc) as a model plant because the underlying mechanisms of metal ion tolerance are well

investigated in this plant (Kra¨mer, 2010). It is known that Zn is largely stored in the vacuoles of N. caerulescens leaf epidermal cells (Cosio et al., 2004; Ma et al., 2005). This may be significantly influenced by the activity of NcMTP1 (van der Zaal et al., 1999; Assunçaõ et al., 2001) in this tissue. However, to our knowledge, nothing is known about differences in membrane protein abundances in epidermal cells in comparison to mesophyll cells. Such differences may be responsible for the preferential Zn transfer into epidermal cells and the mechanisms that enable these cells to cope with the extraordinary high Zn concentrations. We chose a proteomics approach in order to obtain insight into differences in the abundance of the metabolically active building blocks in the two investigated tissues, namely the proteins. The differences in protein abundances resemble differences in metabolic activity more closely than gene transcript levels do. To investigate the distribution of proteins involved in metal hyperaccumulation, differences in the proteomes of leaf epidermis and mesophyll tissue were assessed using a state-of-the-art two-dimensional liquid chromatography/ mass spectrometry (2D-LC-MS/MS) proteomics approach. Quantification of protein abundances was performed based on spectral counting. Furthermore, the Zn speciation in planta was investigated by size-exclusion chromatography/inductively coupled plasma mass spectrometry (SEC-ICP-MS) in both tissue types, in order to identify the dominant Zn species and unravel the processes of Zn accumulation and detoxification in the leaves of N. caerulescens. RESULTS AND DISCUSSION Tissue separation and localization of Zn and marker proteins In order to investigate Zn hyperaccumulation in N. caerulescens, plants were grown hydroponically for 10 weeks and treated with 500 lM Zn for 1 week prior to harvest. For comparison of leaf epidermal and mesophyll tissue, the lower leaf epidermis was peeled off with a forceps and collected for further analysis. We did not succeed in removing and collecting the upper epidermis. Thus, the upper epidermis was removed by sandpaper treatment followed by washing of the remaining leaf tissue with de-ionized water to obtain an epidermis-free mesophyll fraction (Figure S1). As expected, the Zn concentration was much higher in the epidermis (Figure 1). This effect was more pronounced in plants that had been transferred to a Zn-containing medium (Figure 1). While the Zn concentration in cell sap of the mesophyll tissue increased 4.5-fold, from 0.2 to 1.0 mM, Zn concentrations in the epidermis increased 20.8fold from 0.5 to 11.1 mM (Figure 1a). Despite this extremely high Zn concentration, Zn-treated plants showed no visible signs of toxicity in the aboveground tissues (Figure 1b).

© 2012 The Authors The Plant Journal © 2012 Blackwell Publishing Ltd, The Plant Journal, (2013), 73, 131–142

Zinc hyperaccumulation in Noccaea caerulescens 133 Figure 1. Zn concentrations in N. caerulescens leaf cell sap after a 7 day treatment with 500 lM Zn in the nutrition solution. (a) Zinc concentrations in leaf mesophyll (Me) and epidermis (Ep) tissue of Zn-treated (+Zn) and control (C) plants. Values are means ± SE (n = 3). (b) Photographs of control and Zn-treated plants after the 7 day treatment.

(a)

A similar distribution of Zn in the leaf epidermis tissue has been reported previously (Ku¨pper et al., 1999; Cosio et al., 2004; Ma et al., 2005). It is well known that the transcript level of many of the genes involved in the Zn hyperaccumulation pathway does not change in response to Zn supply, as their expression level is constitutively higher in hyperaccumulator species compared to non-hyperaccumulating species (Dra¨ger et al., 2004; Hammond et al., 2006; van de Mortel et al., 2006, 2008; Talke et al., 2006; Weber et al., 2006). To determine whether the gene expression was altered in the plant material used in response to the applied Zn treatment, we used a quantitative PCR approach. Hence, we chose two candidate genes, NcMTP1 (Genbank accession number AY999083) and the leaf nicotianamine synthase gene NcNAS1 (Genbank accession number AJ300446; Mari et al., 2006). A constitutively higher expression rate in leaves has previously been shown in N. caerulescens for both of these genes (Assunçaõ et al., 2001; Weber et al., 2004; van de Mortel et al., 2006; Talke et al., 2006). Our results clearly confirm an expression pattern that was independent of the Zn supply (Figure S2). Based on these findings, we focused on analysis of the protein levels of Zn-treated plant material in order to investigate the distribution of candidate proteins involved in Zn hyperaccumulation in the two tissue types. We performed a comparative proteome analysis of the epidermal and mesophyll tissues of N. caerulescens leaves. In order to compare protein abundances, a spectral counting approach was used. This method is based on the fact that more spectra are assigned to a protein if its abundance is higher. This approach has a high dynamic range, allowing identification of small to very large differences in protein abundances (Bantscheff et al., 2007). Differences in the abundance of proteins involved in Zn hyperaccumulation Total protein fractions as well as proteins from microsomal fractions from epidermal and mesophyll leaf tissues were compared in order to find differences in protein abundances between the two tissue types. In total, 1017 proteins were identified. A list of all identified proteins with

(b)

their identification parameters and abundance in mesophyll and epidermal tissue is provided in Table S1. The complete Mascot MS/MS search results, including the spectral information, are available at http://www.ebi.ac.uk/ pride/ (accession number 21354). Comparison with the distribution of marker proteins (Figure 2) from a transcriptomics approach, comparing the distribution of gene expression in stem epidermis and total stem (Suh et al., 2005), showed that we succeeded in enrichment of mesophyll and epidermal tissue. Proteins located in the chloroplast that are part of the photosystem reaction centers, such as light-harvesting complexes or chloroplastic ATPase, were chosen as markers for the mesophyll tissue (Figure 2, proteins 1–33). As expected, most of these proteins were more abundant in the mesophyll (Figure 2). Likewise, proteins involved in the synthesis and transport of long chain fatty acids (Figure 2, proteins 34–44) were more abundant in the epidermis tissue (Figure 2). Nevertheless, there was a relatively high contamination of epidermal tissue with mesophyll marker proteins. This may be explained by the facts that: (i) guard cells in the epidermal tissue contain chloroplasts that are detected even though the guard cells account for only a small percentage of the total number of cells in this tissue, (ii) the isolated epidermal layer is contaminated with mesophyll cells because part of the vascular tissue is strongly associated with the epidermal cell layer, that is vascular tissue, including chloroplast-containing cells, is still attached to the epidermal cell layer after peeling (Figure S3). Epidermal cells contain far fewer proteins than mesophyll cells because chloroplasts are absent, and because the vacuole (with only very few proteins) makes up a much larger volume (Dietz et al., 1992). Hence, these minor mesophyll contaminations result in relatively high spectral counts from mesophyll marker proteins. Several proteins involved in stress protection, metal transport and metal chelation were differentially expressed in the two tissue types (Figure 2, proteins 45–86). Zinc itself is not a redox-active metal, but it is known that excess Zn may lead to oxidative damage (Weckx and Clijsters, 1997; Cuypers et al., 2001). Thus, cells exposed to high Zn concentrations respond to the resulting oxidative stress by


134 Thomas Schneider et al.

Figure 2. Relative abundances of mesophyll (blue diamonds) and epidermal (red triangles) marker proteins and proteins potentially involved in metal-stress tolerance (green circles), in the S-adenenosyl methionine pathway, synthesis of carboxylic acids, metal transport and metal binding in leaf mesophyll and epidermis tissues of N. caerulescens plants after Zn treatment. To assign all identified hits functionally, protein sequences were compared with the TAIR10 database using the BLAST standalone tool version 2.2.21. Relative abundances were calculated from the number of unique spectra calculated using the Scaffold software: ‘1’ indicates exclusive localization in the mesophyll, and ‘0’ indicates exclusive localization in the epidermal tissue. The blue line shows the mean relative abundance of mesophyll marker proteins. The red line shows the mean relative abundance of epidermal marker proteins. Proteins were assigned as significantly enriched in one fraction when the relative abundance was in a window of mean ± 1 SD of the mesophyll marker proteins (blue area) or epidermal marker proteins (red area). Proteins are sorted according to their relative abundance in the mesophyll fraction. Further details of the identified proteins are given in Table S2.

inducing various anti-oxidative defense mechanisms. We found that glutathione S-transferases (GSTs, proteins 45– 50) were generally more abundant in the epidermal tissue of the N. caerulescens leaves, and proteins 47–50 are significantly enriched in the epidermal tissue. GSTs belong to a large gene family with several sub-classes (Dixon et al., 2002). They function as transferases in the detoxification of xenobiotics, and as glutathione-dependent peroxidases or reductases that participate in detoxification of reactive oxygen species (ROS; Dixon et al., 2009). Their expression is stimulated by salicylic acid, H2O2 (Sappl et al., 2009) and other environmental factors, such as metals, for example Cd (Roth et al., 2006; van de Mortel et al., 2008). Although these studies were performed on root material, Hammond et al. (2006) showed that GSTs are also more abundant in the shoot of the hyperaccumulator N. caerulescens compared to the non-accumulator Noccaea arvense. Furthermore, two cysteine synthase proteins were identified (Figure 2, proteins 59 and 60), one of which (60) showed elevated abundance in the epidermal tissue. Cysteine is required for synthesis of glutathione, which is a

compound that is present at low millimolar concentrations in all plant cells. It has been shown that glutathione levels increase after Ni and Cd supply to hyperaccumulator plants (Freeman et al., 2004; van de Mortel et al., 2008), but there is also evidence against direct involvement of glutathione in hyperaccumulation: when the glutathione biosynthetic pathway in N. caerulescens was blocked, no reduction of Zn hypertolerance was observed (Schat et al., 2002). Our results indicate that epidermal cells are more resistant to Zn-induced oxidative stress than mesophyll cells, as they contain higher levels of GST and possibly also glutathione. Another protein family of interest is the protein disulfide isomerases (PDIs), as two members of this family, PDI1 and PDI2, were more abundant in shoot tissue of the hyperaccumulator A. halleri compared to A. thaliana (Talke et al., 2006). In the present study, we found protein hits from homologs of five A. thaliana PDIs (Figure 2, proteins 51–55), three of which (53–55) were significantly more abundant in the epidermal tissue than in the mesophyll tissue. PDIs have multiple functions in


Zinc hyperaccumulation in Noccaea caerulescens 135 plants, including the ability to bind and detoxify metals such as Cu (Narindrasorasak et al., 2003), as well as Cd and Zn (Rensing et al., 1997). Transport of Zn in the epidermal tissue In order to identify differences in Zn transport capacity into the epidermal cells, protein abundances from microsomal fractions of epidermal and mesophyll tissue were compared. The transporter showing the greatest difference in abundance between the two tissues was a P-type ATPase homolog of AtHMA4 (Figure 2, protein 80), which was 2.7-fold more abundant in the epidermis. In contrast to the situation in A. halleri, in which HMA4 is confined to the plasma membrane of xylem parenchyma cells (Mills et al., 2003; Hussain et al., 2004; Verret et al., 2004; Hanikenne et al., 2008), its localization in the whole leaf blade including epidermal cells has been shown for N. caerulescens (O’Lochlainn et al., 2011). Its significant enrichment in the epidermis indicates that the protein may act as a main player in localization of Zn to the epidermal layer. HMA4 expression has previously been detected in roots and shoots of the Zn hyperaccumulators A. halleri and N. caerulescens (Bernard et al., 2004; Papoyan and Kochian, 2004; Hammond et al., 2006; van de Mortel et al., 2006; Talke et al., 2006). In root tissue, the function of HMA4 is associated with transfer of Zn from the root to the shoot (Hanikenne and Nouet, 2011). In leaf tissue, HMA4 may be responsible for supply of Zn from the xylem and mesophyll cells to the surrounding tissue. Hanikenne et al. (2008) showed that HMA4 expression is detectable in vascular tissue adjacent to the epidermal cell layers in the leaves of A. halleri. By looking more closely at other over-represented zinc transporters localized to the plasma membrane of the epidermal tissue, we identified a protein that showed strong homology to AtZIP4, a member of the zinc/iron-regulated transporter protein (ZIP) family. This transporter was present in the epidermis but not in the mesophyll, where its abundance was below the detection limit (Figure 2, protein 83). For several ZIP family members, higher expression rates were observed in N. caerulescens roots and shoots compared to non-hyperaccumulating relatives (Hammond et al., 2006; van de Mortel et al., 2006). ZIP transporters are well described with respect to Zn uptake from the soil into the root (Verbruggen et al., 2009; Kra¨mer, 2010). However, it is likely that the identified ZIP is involved in Zn uptake in leaf cells. After Zn uptake into the cell, it must be detoxified in order to prevent damaging effects in the cytoplasm. One possibility is sequestration of Zn into the vacuole (Verbruggen et al., 2009). For transport of Zn into the vacuoles, two protein candidates have been described: (i) MTP1, a member of the cation diffusion facilitator (CDF) family, (ii) the P-type ATPase HMA3. We identified NcMTP1 (Figure 2, protein 79), a heavy metal transporter of N. caerulescens

that is a homolog of AtZAT/MTP1 (Kobae et al., 2004; Desbrosses-Fonrouge et al., 2005). This transporter was found to be more abundant in the epidermal tissue but was not significantly enriched in the epidermis. Our findings are similar to those of Ku¨pper and Kochian (2010), who described a similar distribution of MTP1 transcripts in young leaves of N. caerulescens plants. In agreement with the findings by Assunçaõ et al. (2001), our results may indicate a central role for this transporter in enhanced vacuolar sequestration of Zn. MTP1 has also been shown to be more highly expressed in Zn hyperaccumulating N. caerulescens compared to non-accumulating relatives (Assunçaõ et al., 2001; van de Mortel et al., 2006). In contrast to MTP1, the P-type ATPase HMA3 was significantly enriched in the mesophyll tissue (Figure 2, protein 71). Our protein localization data are in agreement with those of Ueno et al. (2011), who showed that NcHMA3 has a higher expression level in the mesophyll. In contrast to HMA3 from A. halleri, which has been shown to have Zn uptake activity (Becher et al., 2004; Ueno et al., 2011), NcHMA3 exhibited no specific Zn transport activity. Instead, it appears that this transporter is more involved in Cd transport into the vacuole (Ueno et al., 2011). Thus, MTP1 may be the only vacuolar Zn transporter in the epidermis cells of N. caerulescens. In addition to the higher abundance of MTP1, the cell sap pH in the epidermal tissue was 0.6 units lower (P < 0.05), corresponding to an approximately fourfold higher proton concentration than in the mesophyll (Figure 4a). Due to the large size of the vacuole, the cell sap pH mainly reflects the vacuolar pH. The lower vacuolar pH in epidermis cells was confirmed in cross-sections stained with the pH-sensitive agent neutral red (Figure 4b). A lower pH represents a stronger driving force for Zn transport via the CDF family member MTP1 than a more alkaline pH as the transporter acts as a Zn/proton antiporter (Kawachi et al., 2008). This has also been described for members of the CDF family from Escherichia coli (Chao and Fu, 2004) and Bacillus subtilis (Guffanti et al., 2002). Zn speciation in mesophyll and epidermis tissue When Zn is transported into epidermal cell vacuoles, it must be detoxified by a Zn-binding ligand in order to prevent export of free Zn ions back to the cytosol. The speciation of the incorporated Zn was investigated by SECICP-MS. Interestingly, our results revealed that, in the mesophyll tissue, almost all of the Zn was strongly complexed to a low-molecular-weight ligand (Figure 3a). This speciation was similar in plants grown under low and high Zn concentrations. However, in the epidermal tissue, only 10% of the total Zn in Zn-treated plants and 25% of the Zn in control plants was present as a stable complex. Further analysis showed that complexed Zn extracted from both tissue types eluted almost exclusively as low-molecularweight complexes (Figure S4a,b). Other elements such as


136 Thomas Schneider et al.

(a)

(b)

Figure 3. Zn speciation in the leaves of control and Zn-treated N. caerulescens plants. (a) Proportion of speciated Zn in epidermis and mesophyll tissues. (b) SEC-ICP-MS chromatogram showing the elution profiles of Zn, sulphur (S) and phosphorus (P) in a Zn-treated epidermis sample.

S and P did not co-elute with Zn (Figure 3b), indicating that S- and P-containing ligands such as glutathione, phytochelatins or phytic acid are not involved in the Zn speciation of N. caerulescens leaves. This is in agreement with the findings of Schat et al. (2002) in Silene vulgaris. In order to further characterize the identity of possible ligands, the low-molecular-weight Zn peak was collected, lyophilized and re-injected onto a hydrophilic interaction liquid chromatography (HILIC) column coupled to an electrospray ionization/time-of-flight mass spectrometer (ESI-TOF-MS). In the peak eluting after 18.75 min, nicotianamine (NA) was identified as the dominant Zn ligand (NA; C12H22N3O6; m/zcalculated 304.1501, m/zanalyzed 304.1598; Figure S4c). Absolute levels of Zn–NA were similar in mesophyll and epidermis (Figure S4a,b), although a much higher proportion of Zn was associated with NA in the mesophyll compared to the epidermis (Figure 3a). NA is mobile in the xylem, phloem and the cytosol, and is an essential compound for Zn, Fe, Cu and Ni transport (Takahashi et al., 2003; Mari et al., 2006; Klatte et al., 2009). It has also been proposed to act as a cytosolic homeostatic buffer, maintaining Zn ions in a non-toxic form. NA forms coordination complexes with high thermodynamic stability for most cationic transition elements, and transport of metal–NA complexes is enabled by members of the Yellow-Stripe-Like (YSL) protein family (DiDonato et al., 2004; Schaaf et al., 2004; Callahan et al., 2006; Gendre et al., 2007). For hyperaccumulating species, higher expression rates of various isoforms of the NA-synthesizing enzyme nicotianamine synthase, together with higher expression of the enzyme S-adenosylmethionine syntethase (SAMS) that synthesizes the NA precursor, have been observed (Weber et al., 2004; Hammond et al., 2006; van de Mortel et al., 2006; Talke et al., 2006). We detected higher expression of SAMS in the epidermal tissue (Figure 2), similar to what was shown by Suh et al. (2005) in a transcriptomics approach using stem material from A. thaliana. This higher epidermal expression did not correspond with

the equal distribution of Zn–NA levels between mesophyll and epidermis (Figure S4a,b). However, it is known that the function of SAMS is not only restricted to the NA biosynthetic pathway. SAMS is up-regulated under various stress conditions, for example salt stress (Espartero et al., 1994) and cold stress (Cui et al., 2005), and its product S-adenosylmethionine is a precursor in many pathways, including ethylene synthesis (Wang et al., 2002), polyamine synthesis (Bouchereau et al., 1999) and synthesis of secondary compounds such as glucosinolates (Grubb and Abel, 2006), which are present in the epidermis of N. caerulescens (Tolra et al., 2001). In conclusion, we suggest that the preferential localization of SAMS to the epidermal tissue is independent of metal hyperaccumulation, and may be more related to the production of plant secondary metabolites. It is well documented that NA is key ligand in Fe, Zn, Ni and Cu homeostasis, ensuring cell-to-cell mobility of these metals (Takahashi et al., 2003; Klatte et al., 2009). A recent study has also shown that the vacuolar-localized ZIF1 protein (Haydon and Cobbett, 2007b), which is required for zinc tolerance, acts as a nicotianamine transporter (Haydon et al., 2012). Our approach extends this model, as, particularly in the mesophyll, the majority rather than the minority of Zn is associated with NA (Figure 3a). As Zn is almost totally localized within the vacuoles of N. caerulescens leaf tissue (Ku¨pper et al., 1999; Ma et al., 2003), NA must also be present in the vacuolar compartment, maintaining Zn in a non-toxic form by chelating the ions in stable complexes. The fact that the absolute amount of NA–Zn complexes was similar in the mesophyll and the epidermal tissue (Figure S4a,b) suggests that NA is not directly involved in the actual Zn hyperaccumulation process in the epidermis; a theory also supported by the findings of Callahan et al. (2007). As only a minor part of the Zn was complexed with NA in the epidermal tissue, further analyses of additional ligands were performed. Organic acids are known to form complexes with several di- and trivalent cations found in plants (Callahan et al., 2006; Verbruggen et al., 2009;


Zinc hyperaccumulation in Noccaea caerulescens 137 Figure 4. Factors potentially involved in metal hyperaccumulation in N. caerulescens leaves. (a) pH of total cell sap in epidermis and mesophyll tissue from control and Zn-treated plants. (b) Micrographs of neutral red-stained leaf cross-sections. Red indicates acid conditions. It is mainly the vacuoles of epidermal and mesophyll cells that are stained red. EC, epidermis cell; MC, mesophyll cell. (c, d) Concentration of (c) citrate and (d) malate concentrations in epidermis and mesophyll tissue from control and Zn-treated plants. Values are means ± SE (n = 3).

(a)

(b)

(c)

(d)

Kra¨mer, 2010). However, compared to NA, the stability constants of these metallo-organic acid complexes are much lower (Callahan et al., 2006), but the low vacuolar pH helps to stabilize the metal–organic acid associations (Haydon and Cobbett, 2007a). For citrate and malate, a constitutively elevated concentration has been observed in various hyperaccumulator species (Lee et al., 1978; Ueno et al., 2005; Montarges-Pelletier et al., 2008). In our study, we found an increased organic acid content in the cell sap of Zn-treated epidermal and mesophyll tissues. Interestingly, consistently higher concentrations were found in the epidermis tissue (Figure 4c,d). Upon Zn treatment, citrate concentration increased from 1.2 to 3.1 mM in the mesophyll and from 2.2 to 3.9 mM in the epidermis. For malate, increases from 19.8 to 39.1 mM and from 38.9 to 76.8 mM were measured in the mesophyll and the epidermis, respectively. The elevated carboxylate levels were in agreement with higher abundances of their synthesizing enzymes, that is citrate synthase, as well as the malate pathway enzymes fumarase, and chloroplastic, cytosolic and mitochondrial malate dehygrogenase and phosphoenolpyruvate carboxylase, in the epidermal tissue (Figure 2). Our results indicate that Zn (11 mM in cell sap) is mainly associated with malate (77 mM in cell sap), and, to a lesser extent, citrate, as the citrate concentration alone (3.9 mM in cell sap) is too low to complex all free Zn ions. These results are in agreement with the situation described in A. halleri shoots (Sarret et al., 2002). Citrate appears to be more important in the process of Ni hyperaccumulation (Lee et al., 1978; Kra¨mer et al., 2000). The SEC-ICP-MS analysis showed that the complexes in the epidermis tissue are labile, which indicates that the dominant binding forms of Zn in the epidermis tissue are weaker than Zn–NA.

CONCLUSION Our results indicate that the leaf epidermal tissue of N. caerulescens is ideally adapted to extreme Zn concentrations. Proteins involved in protection against oxidative stress caused by excess Zn are more abundant in the epidermis than in the mesophyll tissue, and this tissue is also equipped with a highly expressed set of transporters that allows efficient uptake of Zn followed by sequestration into the vacuoles. In the mesophyll, on the other hand, Zn is almost completely complexed with NA. This is important, as these cells are more metabolically active than epidermal cells. They require a mobile yet safe source of Zn, which is controlled by Zn speciation with NA. In the epidermis tissue, the preferred strategy is to deposit Zn in the vacuole where it forms labile complexes with malate and to a lesser degree with citrate. Furthermore, the vacuolar pH is lower in the epidermis than in the mesophyll, which makes vacuolar Zn uptake capacity via MTP1 more efficient. In conclusion, our results show that NA is not only important for cell-to-cell transport, but it also is the preferred Zn ligand in the metabolically active mesophyll cells. In contrast, the metabolically inactive epidermis cells are capable of storing large amounts of Zn in their vacuoles together with the organic acids malate and citrate. The low pH inside the vacuolar compartment may provide a favorable environment for chelating Zn by malate and citrate, as it stabilizes these complexes. EXPERIMENTAL PROCEDURES Plant growth and Zn treatment Noccaea caerulescens var. Ganges seeds were purchased from Guy Delmot. Seeds were sown on aluminum oxide (OIL DRI


138 Thomas Schneider et al. US-special Type III B, Damolin, www.damolin.de), wetted with water, and vernalized for 4 days at 4°C. Afterwards, pots were transferred to a greenhouse chamber, and the plants were grown for 3 weeks at 22°C and 60% humidity. Then they were transferred to hydroponic culture for another 8 weeks, for which the nutrient solution contained 0.43 mM (NH4)H2PO4, 0.3 mM Ca(NO3)2, 2.0 mM KNO3, 10 lM Fe(III)-EDTA, 4.6 lM H3BO3, 0.032 lM CuSO4, 0.011 lM MoO3, 0.076 lM ZnSO4, 0.507 lM MnCl2 and 97.8 lM MgSO4. The nutrient solution was replaced weekly. For Zn treatment, the nutrient solution was supplemented with 0.5 mM Zn(NO3)2 at 1 week prior to harvest.

Tissue collection and protein isolation The leaves were harvested and separated into epidermal and mesophyll fractions. The lower epidermis was removed from the leaf using a forceps, and the tissue was immediately frozen in liquid nitrogen. The upper epidermis was removed by gentle sandpaper treatment of the leaf surface, followed by washing of the remaining mesophyll tissue with deionized water to reduce contamination from the epidermis tissue. Then the mesophyll tissue was frozen in liquid nitrogen. Both tissue types were stored at 80°C until further analysis. Total protein fractions or membrane proteins from microsomes were isolated from both the epidermis and the mesophyll tissue. In order to isolate total protein fractions, the tissues were first ground in liquid nitrogen. Buffer A [50 mM Tris/HCl, pH 7.2, 0.5% SDS, 1 mM CaCl2, 1 mM dithiothreitol and 1 tablet of EDTA-free Complete mini protease inhibitor (1:30 w/v ratio; Roche, www. roche.ch)] was added in a 1:30 w/v ratio. Samples were centrifuged at 3000 g for 5 min at 4°C to remove cell debris. The supernatants were then subjected to overnight precipitation with 10% trichloroacetic acid (TCA) in 25% acetone at 4°C. Proteins were collected by centrifugation at 15 000 g for 30 min at 4°C. The resulting pellet was washed in ice-cold acetone and collected by centrifugation at 15 000 g for 30 min at 4°C. Finally, the pellet was resuspended in 1 ml of 0.5 M triethylammonium bicarbonate/0.1% SDS. To isolate membrane proteins, microsomes were prepared from the epidermal and mesophyll tissues. The collected material was ground in liquid nitrogen and resuspended in buffer B (0.25 M sorbitol, 25 mM Hepes/KOH pH 7.2, 1 mM CaCl2, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride and 1 tablet of EDTA-free Complete mini protease inhibitor (1:25 w/v ratio; Roche). Samples were then centrifuged at 750 g for 3 min at 4°C to remove cell debris. The resulting supernatants were subjected to ultracentrifugation at 100 000 g for 45 min at 4°C. The pellet was washed twice for 1 min in 0.3 M NaI/25 mM Hepes/KOH (pH 7.2) in order to remove proteins attached to membranes, followed by ultracentrifugation at 100 000 g for 30 min at 4°C. Finally, the pellet was resuspended in 25 mM Hepes/KOH (pH 7.2), followed by ultracentrifugation at 100 000 g for 30 min at 4°C. The resulting pellet was resuspended in triethylammonium bicarbonate/0.1% SDS. For all resulting fractions, the protein concentration was determined by the Bradford assay (Bio-Rad, www.bio-rad.com) according to the manufacturer’s instructions. All samples were stored at 20°C until further processing.

Mass spectrometry

com) was added at a 1:50 w/v ratio, and the proteins were digested overnight at 30°C. The peptide mixture was dried in a vacuum concentrator (Eppendorf, www.eppendorf.com) and resuspended in solvent A (10 mM KH2PO4 pH 2.9 in 25% acetonitrile). Peptides were loaded on a 2.1 mm inner diameter 9 200 mm long SCX column (column material poly-sulphoethyl A, 5 lm particle size, 200 A˚, Poly-LC, www.polylc.com) and eluted at a flow rate of 0.2 ml min 1 using solvent B (10 mM KH2PO4, pH 2.9, 0.5 M KCl in 25% acetonitrile). Separation was performed with the following gradient: 0–10 min, 0% B; 10–15 min, 0–5% B; 5– 50 min, 5–35% B; 50–60 min, 35–100% B. The 27 fractions obtained were pooled to seven master fractions according to the SCX spectrum, and desalted using ZipTips (Millipore, www.millipore.com). The samples were analyzed on an LTQ Orbitrap mass spectrometer (Thermo Fischer Scientific, www.thermoscientific.com) coupled to an Eksigent Nano HPLC system (Eksigent Technologies, www. eksigent.com). The solvent composition of buffer A was 0.2% formic acid in 3% acetonitrile, and buffer B contained 0.2% formic acid and 80% acetonitrile. Samples were dissolved in 5% acetonitrile and 0.2% formic acid. Peptides were loaded onto a self-made tip column (75 lm inner diameter 9 80 mm long) packed with reverse-phase C18 material (AQ, particle size 3 lm, 200 A˚; Bischoff GmbH, www.bischoff-chrom.com) and eluted at a flow rate of 200 nl min 1. The following LC gradient was applied: 0–5 min, 3% B; 5–10 min, 3–10% B; 10–55 min, 10–43% B; 55–58 min, 43–97% B; 58–65 min, 97% B. Mass spectra were acquired in the m/z range 300–2000 in the Orbitrap mass analyzer at a resolution of 60 000. MS/MS spectra were recorded in a data-dependent manner for the six most intense signals in the ion trap. Precursor masses already selected for MS/MS were excluded from further selection for 90 sec, and the exclusion window was set to 20 ppm.

Database searches The Mascot search engine (www.matrixscience.com, version 2.3) was used for database searches. MS and MS/MS data were searched against a database containing protein sequences from A. thaliana (TAIR9, 27 379 proteins, downloaded from www.arabidopsis.org), A. lyrata (32 549 proteins, downloaded from http:// www.ncbi.nlm.nih.gov/bioproject/49545) and N. caerulescens (105 protein sequences, downloaded from www.ncbi.nlm.nih.gov). Furthermore, 4289 N. caerulescens ESTs generated by Rigola et al. (2006) were downloaded from www.ncbi.nlm.nih.gov and translated into protein sequences using the OrfPredictor online tool (http://proteomics.ysu.edu/tools/OrfPredictor.html). From that set of sequences, 4157 protein sequences were added to the database. Common contaminants such as keratin and trypsin were also added. The following search parameters were applied: (i) trypsin was used as the protein-digesting enzyme, and up to two missed cleavages were tolerated, (ii) carbamidomethylation of cysteine was chosen as a fixed modification, (iii) oxidation of methionine was chosen as a variable modification. Searches were performed with a parent-ion mass tolerance of ±5 ppm and a fragment-ion mass tolerance of ±0.8 Da.

Data processing version 3.0 (Proteome Software, www.proteomesoftware. com) was used to validate and quantify MS/MS-based peptide and protein identifications. Peptide identifications were accepted if they were established at >95% probability as specified by the Peptide Prophet algorithm (Keller et al., 2002). Protein identifications were accepted if they were established at >90% probability and at least one peptide with a minimum length of nine amino acids, a parent-ion mass tolerance below 3.5 ppm and a Mascot ion score

SCAFFOLD

In order to analyze the proteome of all isolated fractions, 100 lg proteins were subjected to trypsin digestion. Proteins were first reduced by 10 mM dithiothreitol in triethylammonium bicarbonate for 30 min at 50°C. Cysteine residues were then alkylated by treatment with 50 mM iodoacetamide for 1 h at room temperature in the dark. Finally, trypsin (sequencing grade; Promega, www.promega.


Zinc hyperaccumulation in Noccaea caerulescens 139 above 20 was uniquely assigned to a protein in a minimum of two samples. Protein probability was assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). Proteins that were identified using the same set of peptides and were not differentiated by the MS/MS analysis were grouped to protein clusters to satisfy the principles of parsimony. To determine the false-discovery rate, searches were performed against a composite version of our reference database, created by concatenating the target protein sequences with reversed sequences (141 029 entries), as described by Elias and Gygi (2007). The number of assigned reversed hits was multiplied by two and divided by the total number of identified proteins. We calculated a false-discovery rate of