Although metals are natural constituents of soils and occur naturally in the environment (Rascio and Navari-Izzo, 2011). Metal pollution is one of the most serious environmental problems aggravated by human activities as mining and smelting, electroplating, gas exhaust, energy, and fuel production, fertilizer, and pesticide application (Alkorta et al., 2004). In high concentrations, essential and non-essential elements like Cd, Zn, Hg, Pb, Ag, As, Al, among others, are strongly toxic (Upadhyaya et al., 2010). Trace metal elements exert different actions on plants, like blocking essential functional groups, displacing other metal ions, or altering the active conformation of biological molecules (Collins and Stotzky, 1989). In fact, Aluminum (Al) represents about 7% of the earth's solid surface weight. Al is not a free element (Bhalerao and Prabhu, 2013) and Al toxicity increases with soil acidity (Jeffrey et al., 1996). The trivalent cation Al³⁺ has been reported as inhibiting root growth and development. This could be due through Al interactions within symplast, plasma membrane, and cell wall (Kochian, 1995).

Arsenic (As) is one of the most toxic element and can be found in every environmental compartment. In soil, As-inorganic form, as oxidized arsenate (As⁵⁺) and reduced arsenite (As³⁺), are generally more toxic than As-organic form (Mateos-Naranjo et al., 2012). It has been known to cause growth inhibition and membrane disintegration (Moreno-Jiménez et al., 2012). Arsenic toxicity is induced by oxidative stress through generation of reactive oxygen species (ROS) and inhibition of antioxidant defenses (Mateos-Naranjo et al., 2012).

However, plants can regulate their metabolism in response to the trace metal element and protect themselves against their toxicity to a certain extent. Understanding plant-metal interactions may help to reduce the risks associated with the introduction of trace metals into the food chain and to address safety issues in the environment.

Studies have been conducted to improve the knowledge on the tolerance mechanism of plants facing elevated trace metals accumulation without major metabolic alterations (Revathi and Subhashree, 2013). Plants have developed highly complex systems to control the uptake, accumulation and detoxification of trace metals (Leitenmaier and Küpper, 2013). In general, these mechanisms range from exclusion, inclusion and accumulation (Raskin and Ensley, 2000). Possible mechanisms that govern metal tolerance in plant cells (Barceló and Poschenrieder, 1992; Sanità di Toppi and Gabbrielli, 1999) are selective exclusion of metal during uptake, metal excretion, metal retention in the roots, specific tolerance of enzymatic systems, immobilization by means of cell wall, and extracellular carbohydrates, complexation by binding low-molecular weight peptides (phytochelatins) or by ligands such as organic acids and amino acids, and finally by compartmentation (Hall, 2002; Carrier et al., 2003). Thus, trace metals may be stored/accumulated either in cell walls (Lozano-Rodriguez et al., 1997; Carrier et al., 2003), cytoplasm (Rauser and Ackerley, 1987; Carrier et al., 2003), or in cell vacuoles (Carrier et al., 2003). Tolerance may also be achieved in some metal-hyperaccumulating plants by root to shoot metal transport, maintaining low metal concentration in roots (Kramer et al., 1997).

Tamarix gallica L. is a halophyte from coastal and desert regions, living also in some saline depressions, which are usually used as sites of accumulation of industrial and urban effluents contaminated by metals (Ghnaya et al., 2007). In order to survive in these sites and according to their tolerance capacity, plants may have mechanisms to regulate internal, and cell wall metal concentrations by which determine their survival (Sousa et al., 2008). The genus Tamarix is a relatively long-living plant that can tolerate a wide range of environmental conditions, resisting abiotic stresses such as salt, high temperature, and drought stresses (Saïdana et al., 2008). Additionally, Tamarix spp. has also been studied as a model for morphological traits for adaptation to saline and contaminated environments (Moreno-Jiménez et al., 2009; Haiyan et al., 2011; Sghaier et al., 2015).

The mechanisms for internal detoxification of As and Al in T. gallica have not been well documented and little known about the localization and speciation of As and Al in this halophyte. To determine the localization and distribution of elements inside plants, different methods can be used, such as histochemical (Vázquez et al., 1992b; Seregin and Ivanov, 1997), cell fractionation (Lozano-Rodriguez et al., 1997), X-ray microanalysis (Khan et al., 1984; Vázquez et al., 1992a,b), particle-induced X-ray emission (micro-PIXE) (Ager et al., 2002), or nuclear micro-probe technique (NMP) (Ager et al., 2003). The aim of the present work is to determine the tissue and cellular localization of As and Al in roots and leaves of T. gallica by using cell fractionation. Knowledge of As and Al distribution in plants can help to clarify the process involved in the uptake, transport, and deposition or detoxification of these elements in T. gallica.

Roots presented significantly higher metal concentrations than the leaves, for all studied treatments (Table 1). Metal translocation from roots to leaves can be expressed by the translocation factor (TF), and varied from 0.192 ± 0.034 to 0.087 ± 0.010 in the case of Al, showing that the treatment with the lowest Al concentration had the highest TF. The opposite pattern was observed for As, with TF varying from 0.071 ± 0.012 to 0.188 ± 0.028, with the highest TF being found in the highest metal concentration supplied with salt (Table 2).

Aluminum was predominantly bound to cellulose in the leaves (Figure 2A), ranging from 43.14 ± 3.69 μg.g⁻¹ to 61.36 ± 6.92 μg.g⁻¹ (Table 1). In the roots, Al was mainly bound to lignin (Figure 2B), with concentrations ranging from 227.45 ± 32.28 μg.g⁻¹ to 510.66 ± 22.14 μg.g⁻¹ (Table 1). Concerning As, higher concentration was found in the polysaccharidic fraction in the leaves (Figure 2C), ranging between 24.55 ± 5.09 μg. g⁻¹ and 90.60 ± 13.75 μg.g⁻¹ (Table 1); in the roots, the difference between fractions was less evident (Figure 2D), and similar concentrations were present in the polysaccharidic, the aqueous, and the proteic fractions (Table 1).

Considering the three fractions in which metal compartmentation can be grouped in this work (cell wall, proteic and intracellular/soluble fractions, Figure 3), 95.17 ± 1.1% of Al in the leaves was accumulated in the cell wall (with total absence in the pectic fraction) and with very low values in the soluble (3.66 ± 1.2%) and in the proteic fraction (1.17 ± 0.9%) (Figures 2A, 3A). The same predominance of Al in the cell wall components (98.3 ± 0.4%) was observed in the roots, with a residual presence of Al bound to proteins (0.5 ± 0.2%), and 1.2 ± 0.3% bound to soluble components (Figures 2B, 3B). Shoots accumulated on average 78.0 ± 2.8% of As into cell wall components (pectin, polysaccharides, lignin and cellulose), 9.9 ± 1.4% was retained in the proteic fraction and 12.1 ± 2.2% in the soluble fraction (Figure 3C). In the roots, 36.0 ± 2.6% of metal, on average, was retained inside the cell, 44.3 ± 2.3% in the cell wall, and the proteic fraction retained 19.7 ± 1.0% (Figure 3D).

In the absence of salt, total concentration of Al in the leaves showed non-significant differences (p > 0.05) with increasing treatment concentrations (Table 1). There were, however, significant differences in Al compartmentation in the leaves, the intracellular / soluble fraction presented significant differences (F = 39.462, p < 0.001), showing a peak accumulation under the 500 μM Al treatment, and with the highest concentration (800 μM Al) resulting in the lowest accumulation in this fraction. Within the soluble fraction, the difference was particularly marked in the aqueous fraction (H = 7.448, p = 0.024), since Al was not detected there when the highest treatment concentration was applied. On the other hand, Al bound to proteins was not detected for the lowest concentration (200 μM Al), but increased with increasing treatment concentrations (F = 366.373, p < 0.001). The pectic fraction did not show any detectable concentration of Al in the leaves for any of the treatments. Unlike what was observed in the leaves, total Al concentration in the roots presented significant differences across treatments (F = 5.656, p = 0.046), with the highest accumulation being verified when 500 μM Al were supplied in the nutritive solution; a corresponding increase was observed in the Al bound to cell wall components (H = 6.489, p = 0.039). In the more detailed partitioning scheme (Figure 2B), it was possible to observe that when the highest concentration of Al was present in the nutritive solution (800 μM), no Al was detected bound to the proteic fraction (H = 6.161, p = 0.046); Al in the ethanolic fraction increased with 500 μM Al in solution, but decreased when the highest concentration was used (Table 1).

With the addition of 200 μM of NaCl, the response of total Al accumulation in the leaves was similar to that without salt, with no significant differences across treatments (p > 0.05). On a closer look, however, some differences were found compared to the “no salt” treatments, Al in the aqueous fraction was only detected in the lowest treatment concentration (200 μM), and the proteic fraction was only present when 800 μM Al was used in the experiment. Parallel to the experiment without NaCl, no Al was detected in the pectic fraction. There was also a displacement of Al toward lignin as more metal was present in the nutritive solution, with significant differences between the two extreme concentrations, 200 μM and 800 μM (F = 14.676, p = 0.005). In the roots, there was an increase in total Al accumulation with increasing treatment concentrations, but differences were only significant between extreme values (H = 6.489, p = 0.039). In the presence of salt there was generally a higher accumulation by T. gallica's roots (Table 1). Cell wall components accumulated more Al with the 800 μM treatment (H = 6.489, p = 0.039), particularly polysaccharides (F = 9.871, p = 0.013) and lignin (F = 26.429, p = 0.001); the pectic fraction showed a peak accumulation from the 200 μM to the 500 μM treatment, but remained stable when more metal was added to the solution (F = 28.1812, p = 0.001). Intracellular/soluble Al was more readily accumulated in the roots at lower concentrations in the presence of salt, but decreased with 800 μM Al in solution (F = 13.915, p = 0.006).

Total concentration of As in the leaves increased significantly with increasing treatment concentrations, both with and without NaCl added (Table 1). In the absence of salt, the difference was only significant between extreme concentrations (F = 7.520, p = 0.023), while with salt added to the nutritive solution the greatest increase was observed in the 500 μM treatment (F = 19.602, p = 0.002). In the absence of salt, the increase in total As corresponded to an increase in the cell wall components (F = 8.215, p = 0.019), specifically in the polysaccharidic fraction (F = 6.906, p = 0.028). With salt added to the treatments, there was also a significant increase in the accumulation of As in cell wall components (F = 14.377, p = 0.005), not only involving the polysaccharides as before (F = 10.623, p = 0.011), but also the pectic fraction (F = 89.366, p = 0.00003). Arsenic was only bound to lignin at the highest concentration of As. Increasing As concentration with NaCl has also promoted significant changes in the soluble fractions (F = 13.915, p = 0.006), with the 800 μM treatment resulting in greater As accumulation in the ethanol fraction (F = 8.370, p = 0.018). As for the roots of T. gallica, no significant differences were observed across treatments with the addition of salt in neither total As accumulation nor in its compartmentation.

Translocation—the process by which metal ions taken up by roots are channeled to the shoots and to the intracellular organelles (Revathi and Subhashree, 2013) was rather low in the present study regarding all treatments under consideration. As and Al concentrations decreased markedly from roots to leaves, suggesting a retention of both elements in their long-distance translocation, which can be considered as the primary tolerance/defense mechanism of T. gallica. Although, As and Al concentrations in the aerial part increased with increasing metal supply, a lower translocation factor was recorded when higher concentrations of both elements were present, implying that As and Al translocations to the aerial part are restricted by internal barriers to defend the above ground part. Metal-tolerant plants accumulate higher concentration of toxic metals in roots and lower in shoots, contrary to hyperaccumulators (Lu et al., 2013). The root system is the dominant metal uptake route, with only small amounts being translocated to aboveground parts (Caçador et al., 2009). Arsenic is known to have low mobility regarding translocation from roots to shoots, except in hyperaccumulators. In wild-type Arabidopsis thaliana, 2.6% of As taken up by the roots was translocate to the shoots (Quaghebeur and Rengel, 2004), and several other studies reported an inefficient As translocation ratio (e.g., Marin et al., 1992; Carbonell-Barrachina et al., 1998; Burló et al., 1999; Pickering et al., 2000). Arsenic limited translocation may result from rapid reduction of arsenate, As(V), to arsenite, As(III), in the roots, followed by complexation with thiols, and possibly sequestration in the root vacuoles (Zhao et al., 2009). When the arsenate reductase AtACR2 in A. thaliana was silenced using RNAi, As accumulation in the shoots increased markedly (Dhankher et al., 2006). It was suggested that blocking AtACR2 would lead to more arsenate in the roots available for xylem transport to the shoots, presumably via the phosphate transport pathway. Metal uptake can also be inhibited by the formation of organic acids-metal complexes (Revathi and Subhashree, 2013), for example, citrate is known to inhibit Al uptake, working as an Al-tolerance mechanism in several plant species (De la Fuente et al., 1997; Papernik et al., 2001; Pineros and Kochian, 2001), and malic acid released from the radicle apex of wheat increases when Al is supplied (Delhaize and Ryan, 1995).

In general, metals are accumulated in the plant in different cell compartments, linked to different cell compounds (Reboreda and Caçador, 2007; Sousa et al., 2008). To reduce metal toxicity to the plant, they are preferentially driven away from metabolic active compartments (such as chloroplast and mitochondria) and are stored/sequestered in vacuoles and in the cell wall (Küpper et al., 2001; Psaras and Manetas, 2001; Sousa et al., 2008). To our knowledge, no other studies of As and Al localization in T. gallica have been undertaken. The present study was carried out to investigate the uptake and the distribution of As and Al in this species, after previous works suggested that T. gallica possesses great ability in tolerance and accumulation for both As and Al (Sghaier et al., 2015).

In T. gallica, As and Al were found to bind to cell wall compounds in different percentage, in roots and leaves. Arsenic presents chemical similarity with P, although its soil chemistry is more diverse, and binds readily with S and C (Wenzel, 2013), while Al has high affinity with oxygen donors, like carboxylate and phosphate (Chang et al., 1999) and organic acids like malate and citrate (Gupta et al., 2013).

Globally, Al compartmentation presented greater consistency between roots and leaves, with most metal bound to the cell wall (>95%) with either absence or presence of NaCl. It has been reported that the diminution of the metal concentration in the cytoplasm could be related to the metal accumulation in cell walls playing a role as a barrier against harmful effects by metals (Ramos et al., 2002; Zornoza et al., 2002). Aluminum forms such strong bounds to the cell wall that its amount will generally remain unaltered with the addition of other metallic cations (Krzesłowska, 2011). Arsenic partitioning resulted differently from what was observed with Al, most As in the leaves was found bound to the cell wall, but the importance of this location showed a slight decrease in the presence of salt. Arsenic compartmentation in the roots was more evenly distributed, with the proteic and soluble fractions containing more than 50% of total As together.

Aluminum is the third most abundant element and the most abundant metal in the earth's crust (Ma, 2005), but the most common forms in which occurs render Al generally unavailable to the plants (Delhaize and Ryan, 1995). However, when available, Al presents high toxicity to plants, namely though its soluble ionic form, Al³⁺ (Ma, 2007), promoting growth impairment, inhibition of root elongation, plasma membrane, and cell wall damages (Ma, 2005 and references therein), as a result of binding to intra- and/or extra-cellular compounds (Ma, 2000). Compartmentation of Al in cell walls is a very important mechanism responsible for Al detoxification, as it was observed in the major Al hyperaccumulator, Camellia sinensis (Gao et al., 2014), in Chara corallina (Tolra et al., 2011), and in cultured tobacco cells, where Al uptake and distribution showed that most Al (>90%) accumulated in the cell wall (Chang et al., 1999). The fact that the main percentage of metals was bound to the cell wall rather than located intracellular may have crucial significance as a detoxifying mechanism in T. gallica leaves and roots. Early works (Matsumoto et al., 1976) explained that Al localization in epidermal cell walls was linked to the movement of the water mass flow from the xylem to the epidermis, where the water evaporates into the air, and therefore Al is concentrated in the cell walls at the sites of evaporation. The limitation of translocation into shoots and metals binding to the cell wall are well known mechanisms for non hyperaccumulator plants (Wagner, 1993; Grant et al., 1998). However, the pattern of compartmentation seems to be species and metal-specific. For instance, heavy metals as Ni, Cd and Zn are generally stored in the vacuoles of epidermis cells (Küpper et al., 1999; Psaras et al., 2000). Aluminum forms strong bonds with oxygen donors, such as phosphates and carboxylates, and has been described as strongly binding to the carboxylate group of pectin in the cell wall (Chang et al., 1999). In the present work Al did not accumulate in the pectic fraction in the leaves, regardless of the treatment applied, and its accumulation in the roots was of minor expression (<3% of total Al, on average). Instead, cellulose was the preferential binding site in the leaves, and lignin, cellulose, and polysaccharide's fractions were the preferred binding locations in the roots. The binding of Al to cell wall pectin may not always be an essential tolerance mechanism (Krzesłowska, 2011). At high external concentrations, the proteic fraction acquired importance in the leaves, particularly in the absence of salt. With NaCl present in the nutritive solution, Al bound to the proteic fraction was only detected at the highest treatment tested. Tolerance mechanisms in plant cells include chelators such as phytochelatins to avoid metal stress into the cytoplasm; these polypeptides can bind metal cations and sequester them in the vacuole, allowing the cells to cope with higher external metal concentrations (Krzesłowska, 2011). At higher concentrations, the roots showed a displacement of Al toward lignin. The increase in cell wall lignification, and subsequent metal deposits therein, is another mechanism that has been described to protect plant cells from toxic effects of high concentrations (Probst et al., 2009).

A large proportion of As was bound to cell wall components in the leaves, which seems to function as the first barrier protecting plant from As toxicity. In lettuce leaves, 64% of total Cd was bound to cell walls (Ramos et al., 2002) and a similar proportion of Cd associated with the cell wall fraction was observed in Lupinus albus (Zornoza et al., 2002). Large portions of metals are accumulated in the cell wall-plasma membrane interfaces and thus it could be considered as the potential tolerance-related site (Hossain et al., 2012). The cell wall played an important role in retaining As in the leaves of T. gallica outside metabolic sites, with over 70% of the As in the root was retained there. It is possible that As forms strong binding to cell wall components like cellulose, pectin and lignin (Koch et al., 1999), which could immobilize excess As in T. gallica leaves. Besides the cell wall, the vacuole seems to be another site of preferential As binding in leaves. In our study, As analysis at the subcellular level of plant tissue demonstrated the importance of the soluble fraction. The vacuole in root cells is most likely a sink for As. Vacuolar compartmentation has been described as playing a very significant role in detoxification and tolerance, which prevents the circulation of free ions in the cytosol and forces them into a limited area (Sanità di Toppi and Gabbrielli, 1999). The proteic fraction has also showed great importance in the compartmentation of As in T. gallica roots. Transport to and storage in the vacuole requires increasingly sulfur-rich peptides, including phytochelatins and organic acids (Sanità di Toppi and Gabbrielli, 1999). Phytochelatins (PC) are synthesized from glutathione by PC-synthases, and play a role in the distribution and accumulation of As and some other highly toxic metals like Ag, Hg, and Cd (Rauser, 1999; Cobbett, 2000) thus removing these toxic elements from the cytosol (Verbruggen et al., 2009). The inactivation of toxic metal ions through the synthesis of phytochelatins and then the formation of metal–phytochelatin complexes is a general mechanism for trace metal homeostasis in plants (Zenk, 1996). It has also been demonstrated that efflux transporter (PvACR3) is oriented by tonoplast, which mediates As(III) sequestration into the vacuoles in hyperaccumulator fern, Pteris vittata (Indriolo et al., 2010). Additionally, Pickering et al. (2000) reported that As(III) in Brassica juncea is coordinated with sulfur groups. Depending to the acidic conditions which confer the stability of such complexes, As-PC complexes could be stored in the vacuoles (Meharg and Hartley-Whitaker, 2002). Further, complexation of metals with organo-ligands within the storage sites results in decreased free ion activity and thus reduced toxicity (Sanità di Toppi and Gabbrielli, 1999; Bhatia et al., 2005). Previous study by X-ray microanalyses, revealed that As was localized in the central intracellular portion of cells. Since the majority of this area is occupied by the vacuole, this suggests that As was mainly contained in the vacuoles (Lombi et al., 2002), which is in accordance with our findings of decreased importance of the cell wall as a defense mechanism to As accumulation. Further, an observation using EDXA suggested that the compartmentation in the vacuoles could be considered as an efficient strategy to maintain a low As concentration in the cytoplasm (Lombi et al., 2002).

Compartmentation is a key aspect in removing trace metal elements outside of key metabolic sites, contributing to the plants survival in salt depressions contaminated by these elements. This study gives an insight into the different distribution and localization of As and Al inside T. gallica. Further analyses to examine the distribution of As and Al and their bonding state at the ultra-structural level may improve our understanding of the effects of both elements on cell ultrastructure within the plant. A better understanding of metal sequestration in plants may eventually contribute toward the development of biorecovery techniques for the remediation of soils contaminated with these two traces metals.

From our study, we deduce that both the vacuole and the cell wall are involved in the As and Al tolerance mechanisms to protect metabolically active cellular compartments from toxic concentrations, and different bonding strategies are clearly involved depending on the element. This should not only be related to metal ligand specificities, but also to the distinct abundance of Al and As in the environment, and their degree of toxicity to the plant. Salt appears to exert some influence on metal compartmentation at a finer level, but overall there are fewer differences regarding the distribution among soluble, proteic, and cell wall bound metal. Thus, it can be stated that compartmentation in T. gallica is metal(loid) and organ specific, and a crucial tolerance mechanism.

DBS and SP carried out all the experiments, assured data analysis and prepared the manuscript. BD and MSD assured samples' analysis. IC and NS conceived the idea, supervised the work and corrected the manuscript. All authors read and approved the final manuscript.