Mechanisms of metal toxicity in plants

Abstract

Metal toxicity in plants is still a global problem for the environment, agriculture and ultimately human health. This review initially addresses the current state of the environmental/agricultural problem, and then discusses in detail the occurrence, mechanisms and relevance of toxicity of selected trace metals (Cd, Cu, Fe, Hg, Ni, and Zn). When discussing the mechanisms, special emphasis is laid on a critical review of their environmental/agricultural relevance, because even now many studies in this field of research are performed under highly artificial lab conditions. The main problems outlined in published studies are artificially high metal concentrations (which never occur even in highly polluted sites) combined with too short treatment times, as well as environmentally and agriculturally irrelevant growth conditions (e.g. constant light and submerged cultivation of seedlings). Furthermore, wherever possible an attempt is made to link the mechanisms published to date in terms of discussing which mechanisms are a direct cause of the observed disturbance of plant function and which are rather a consequence of the primary mechanisms, leading to a complicated toxicity phenotype and ultimately to diminished growth or even death of the plants.

Graphical Abstract

Metal toxicity in plants is still a global problem for the environment, agriculture and ultimately human health.

Introduction: environmental relevance of metal toxicity in plants

Metal toxicity is still a global problem, although public perception is different. After strong efforts towards improving water purification plants had been made, the “green” movement started in the 1970s–1990s. In many “Western” countries, this problem is now regarded as being a problem of the past, which is now only relevant in “developing” and “transition economy” countries. This is, however, a wrong conclusion, because even in the richest countries of the “Western” world, metal toxicity is still a problem for plants and the environment in general. This is made obvious by recent data, e.g. that published by the European Environment Agency on the development of cadmium and mercury discharges into the North Sea, where a rise can be observed since the mid-1990s. A similar trend can be observed in data of cadmium emissions into rivers.¹ This contamination originates from many sources. A selection of toxic metal concentrations in urban soils, unpolluted and polluted soil environments, together with guide values from the US and European Environmental Protection Agencies is listed in Table 1; respective data for aquatic environments is listed in Table 2. Different soil types and pHs can strongly influence the behaviour of metals in soil and thereby the bioavailability of metals,² thus these have to be considered.

Table 1

Average metal concentrations in mg kg⁻¹ in urban soils, contaminated areas and guide values for soil-clean ups

City/region	Cd	Cu	Fe	Hg	Ni	Zn	Ref.
Pittsburgh, USA	1.2	—	—	0.51	—	—	Carey 1980233
La Coruña, Spain	0.3	60	—	—	28	206	Cal-Prieto 2001234
Madrid, Spain	—	71.7	—	—	141	210	De Miguel 1998235
Aberdeen, Scotland (parkland soils)	—	27	18 469	—	14.9	58.4	Paterson 1996236
Aberdeen, Scotland (roadside soils)	—	44.6	18 116	—	15.9	113	Paterson 1996236
Hong Kong	2.18	24.8	—	—	—	168	Li 2001237
Palermo, Italy	0.82	75.5	—	1.85	18.8	149	Manta 2002238
Montreal Island, Canada, historic industry area, 3 rail yards	2.3–7.3	160–245	—	—	64–98	410–547	Ge 2000239
Baltimore, USA, historic industry area	1.06	45	23 495	—	27	141	Yesilonis 2008240
Guizhou, China (mining area) at smelter	60.5	202	—	—	24.1	2551	Li 2007241
Guizhou, China (mining area) 15 km away	5.1	72.6	—	—	9.9	867	Li 2007241
Sicily, Italy, unpolluted	1.3	34	—	0.066	—	122	Modified from Manta 2001238
New York State Department of Environmental Conservation max. values for unrestricted use (incl. agriculture)	0.43	270	—	0.81	72	1100	NYS DEC242
NYS DEC max. values for residential use	0.86	270	—	0.81	140	2200	NYS DEC242
Quebec guidelines for soil cleanup: clean	1.5	50	—	—	50	100	Ge 2000239
Quebec guidelines for soil cleanup: should be restored	5	100	—	—	500	500	Ge 2000239
Quebec guidelines for soil cleanup: needs immediate cleaning	20	500	—	—	1000	1500	Ge 2000239

Table 2

Averages (or ranges) of metal concentrations in nM in freshwater environments

Region	State of contamination	Cd	Cu	Fe	Hg	Ni	Zn	Ref.
Lakes in north UK	Unpolluted	0.23	7.5	1251	—	9.6	56.7	UK EPA, 2008243
Lake Constance, Germany	Unpolluted	<0.44	<7.9	134	<0.25	8.5	—	Lake Constance, Zvbwv.de244
Swedish Lakes, Sweden	Unpolluted	0.044–0.14	4.8–7.9	—	0.005–0.02	3.4–6.8	13.6–30.3	Swedish EPA245
Swedish streams, Sweden	Unpolluted	0.027–0.14	14.3–30.2	—	0.005–0.02	8.52–46	44–86.4	Swedish EPA245
Stream waters Ontario, Canada	Polluted	10.4	30.3–59.8	2059–8219	0.2	26.4	181–217	Ontario monitoring, online246
3 lakes in mining area in Ontario, Canada	Polluted	—	80–254	—	500–20 000	170–1703	—	Ontario Water quality Report 2012247
Lakes in North UK	Polluted	23.9	42.3	—	—	91.4	23 897	UK EPA243
Animas River, Colorado after Gold mine spill 2015	Polluted	21.4	857	3760	0.4	90	4545	US EPA report, Gold mine response248
Water quality criteria for aquatic life – acute	—	17.8	—	—	6.98	8007	1818	US EPA249
Water quality criteria for aquatic life – chronic	—	2.2	—	17 907	3.84	886	1818	US EPA249

In many cases, agriculture itself is a major source of both cadmium and copper contamination that can ultimately threaten agriculture and human health.³ A well-known but nevertheless still unsolved problem, is that copper compounds are used as pesticides in vineyards, which can lead to micromolar copper concentrations in agricultural field runoff and small creeks nearby.⁴ Such concentrations become lethal to sensitive aquatic plants within a few days; toxicity can occur with concentrations as low as 20 nM copper.⁵ Furthermore, the similarly well-known but unresolved problem of cadmium contamination of phosphate fertilisers leads to cadmium contamination of fields that have been used for intensive agriculture for a long time.

In this latter case, strategies for improved phosphorus utilization, dramatically reducing the need for phosphate fertilisation, have been developed but are still too rarely used due to a lack of public awareness of the problem. Another strategy to decrease heavy metal toxicity stress and heavy metal accumulation in crop plants is intelligent breeding based on recent insights into mechanisms of metal uptake, translocation, sequestration and storage, as described in the review by Khan et al. (2014).⁶

Regarding metals, plants are divided into three groups: “Excluders” actively remove excess metals from their tissues, resulting in constant concentrations in the shoot over a wide range of metals in the soils. “Indicators” have no avoidance strategy and accumulate increasing amounts of metals in proportion to increasing concentrations in the soils. “Hyperaccumulators” actively accumulate metals in their shoots, with high bioaccumulation coefficients at low soil concentrations, partially saturating at higher soil concentrations.^7–9

Toxicity mechanisms are best investigated using indicator plants, although comparisons of hyperaccumulating and non-hyperaccumulating plants from the same species can yield valuable insights into their respective toxicity or detoxification strategies.^10,11 Furthermore, plants that are hypertolerant hyperaccumulators for some metals may be indicators to others, which they do not hyperaccumulate. This applies e.g. to the Cd/Zn hyperaccumulator, Noccaea (formerly Thlaspi) caerulescens, which tolerates Cd and Zn in very high concentrations¹² but experiences Cu toxicity like other non-accumulating plants.¹³

Plants are immobile and thus cannot move away from unfavourable conditions, such as toxicity or deficiency of certain elements. In this review, we focus on the biophysical and biochemical mechanisms of metal toxicity in plants and algae for six selected metals (Cd, Cu, Fe, Hg, Ni and Zn).^† Excellent reviews regarding other toxic metal(loid)s like As,^14–16 Al,^17,18 Cr^19,20 or Pb^21,22 can be found elsewhere.

In this review, we will critically analyse suggested mechanisms of metal toxicity in view of the fact that many studies were not performed under environmentally relevant conditions. For example, studies using constant light²³ are questionable for the following reasons. In nature, plants and algae always experience changing light intensities due to clouds or turbid waters and they have developed strategies to optimize photosynthesis and minimize photodamage.²⁴ However, something they never experience under environmentally relevant conditions is constant light. Without a dark phase, plants and algae experience light stress and can be severely injured (see review by Velez-Ramirez et al., 2011)²⁵ and, for example, cannot synchronize cell division to the light cycle.²⁶ Another problem linked to investigation of toxicity mechanisms is the mode of exposure to the toxic metals. Exposure of leaf discs,^27,28 callus cultures²⁹ or submerged seedlings^30,31 to heavy metals can only give some indications on ways in which metal stress would impact the whole plant. Signalling molecules may be lacking or toxicity might occur in the roots, which cannot be detected using leaf discs. Furthermore, submerged seedlings may suffer much more from oxygen and carbon deprivation due to the lower gas diffusion rates in water compared to air. All these artificial stresses could be stronger than the metal stress that was the subject of investigation (reviewed by Mommer & Visser, 2005 and Voeseneck et al., 2006).^32,33 The third and still most common problem in metal toxicity studies is the use of artificially high metal concentrations, which would almost never occur even in the most polluted environments. This often leads to results that are irrelevant to the mechanisms of toxicity in the environment, because at very high concentrations of a particular metal, inhibition becomes unspecific. This is caused by the chemically well-known fact that once all high-affinity binding sites are occupied, metal binding will start to occur at low-affinity sites. Important examples of such cases will be discussed in detail in the following sections of this review.

In this review, we generally avoided the issue of metal deficiency except where needed for understanding toxicity. Resistance mechanisms (e.g. gene regulation, metal detoxification, sequestration and lignification) were included where they overlapped with toxicity mechanisms. We summarized the findings presented in the text in a scheme (Fig. 1). Some metals induce toxicity by similar mechanisms, which are usually described for only one metal in detail in the following sections.

Fig. 1

Mechanisms (including interactions) of heavy metal toxicity in plants as described in this review. The different sizes of the elemental symbols refer to a more or less pronounced effect of that metal.

Toxicity of biologically redox-inert metal ions

Since for many years oxidative stress led news headlines about metal toxicity, many readers might assume that this is the main cause of heavy metal toxicity. While redox reactions do play a role in the development of stress symptoms and there are distinct differences between individual metals as discussed below, oxidative stress is often not the main cause but rather a consequence of metal toxicity. Keeping this in mind, it is not surprising that (a) biologically redox-inert metals also ultimately cause oxidative stress whenever they become toxic and (b) probably the most toxic metal, mercury, is usually redox-inert in plants.

Zinc – an essential element with low but relevant toxicity

Zinc is, in many parts of the world, more a problem in terms of deficiency than toxicity^34,35 and Zn²⁺ is far less toxic to most plants compared to, e.g., Cd²⁺ at the same concentration. The New York Brownfield directive (NYS DEC) for soil clean-ups allows unrestricted use (incl. agricultural) of soils with 1100 mg kg⁻¹ Zn²⁺ (see Table 1). Nevertheless, both naturally Zn²⁺-rich sites and anthropogenically Zn²⁺-contaminated sites exist, on which Zn²⁺ toxicity limits plant growth and makes agriculture impossible.

Zn toxicity originates, to a large extent, from replacement of other weakly-bound divalent metal ions from essential sites. One such site is that of Mg²⁺ in chlorophyll (Chl). This Mg-substitution, when occurring in an uncontrolled way in a system that evolved for [Mg]-Chl, inhibits photosynthesis for several reasons. First, [Zn]-Chl has, like all other heavy metal-substituted Chls ([hms]-Chls), a less stable singlet excited state.³⁶ Therefore, electrons from the excited antenna have a reduced likelihood of being transferred to the reaction centre to perform charge separation and a higher likelihood of being dissipated as heat instead. Furthermore, like all other [hms]-Chls, [Zn]-Chl has a diminished tendency to bind axial ligands. Since these are essential for proper binding in Chl proteins and the tertiary structure of these proteins is only stable with bound Chls,³⁷ Mg-substitution leads to denaturation of pigment–protein complexes.³⁸ In environments where cells constantly have to cope with high levels of heavy metals (incl. zinc) and additionally with very acidic conditions, bacteria of the genus Acidiphilium have evolved that perform photosynthesis with [Zn]-BChl.³⁹ It seems that in these organisms, the disadvantages of [Zn]-BChl compared to [Mg]-BChl are over-compensated by the much higher stability of the Zn-complex. This higher stability probably prevents demetallation or uncontrolled transmetallation of their chlorophylls, which would most probably occur if these organisms used regular [Mg]-BChl. The substitution of Mg²⁺ in Chl leads to the degradation of whole photosystems. The loss of Chl and other photosynthetic pigments is the cause of visible chlorosis, which is a typical visible symptom of Zn²⁺ toxicity, besides reduced growth, leaf necrosis and also reddening of leaves due to anthocyanin production.^40,41 In sugar beet, moderate Zn²⁺ toxicity (50 μM) reduced all photosynthetic pigments and iron (Fe) content, while Chl fluorescence parameters and gas exchange did not change.⁴² Comparable results were found in Fe-deficient sugar beet,⁴³ suggesting Zn-triggered Fe deficiency as the main stressor. Iron is needed in several Fe–S clusters in photosynthesis and respiration; therefore, one would expect a change in Chl fluorescence and gas exchange. Higher Zn²⁺ concentrations (100–300 μM) led to significant inhibition of photosynthesis, which was not related to Fe deficiency. Although using extremely high Zn²⁺ concentrations (1–50 mM, in soil), Bonnet et al. noticed a decrease in F_v/F_m in ryegrass, indicating loss of functional photosystems.⁴⁴ Chlorophyll a fluorescence kinetics give information about the photosynthetic performance of a plant and F_v/F_m is the most frequently used parameter. It measures the maximal dark-adapted photochemical quantum yield of photosystem II (e.g. review by Maxwell and Johnson, 2000).⁴⁵ However, ryegrass can seemingly endure high Zn²⁺ concentrations before the onset of toxicity symptoms, although not all the Zn will be bioavailable for the plants.² Unfortunately, the authors did not determine the bioavailable (or at least labile) Zn content in the soil.⁴⁴ A noticeable decrease in F_v/F_m in plants treated with 1 mM occurred after only 20 days of exposure.⁴⁴ No changes in F_v/F_m or the Chl content were found in Myracrodruon urundeuva plants exposed to Zn²⁺ concentrations of up to 200 mg kg⁻¹ in soil, although phytotoxicity symptoms occurred. Increased carotenoids and antioxidants may have prevented toxicity to the photosystems.⁴⁶ Zinc-containing (0.5 mM) wastewater from an electroplating unit strongly inhibited PS-II-mediated electron transport and lowered F_v/F_m, as well as lowering the activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCo; see also below). In contrast, PSI-mediated photoreactions increased (both measured on isolated thylakoids), indicating that enhanced cyclic electron flow dissipated excitons.⁴⁷ Inhibited electron transport activity of isolated thylakoid membranes from Zn²⁺-treated maize plantlets suggested an inhibition of the water oxidation complex (possibly by substitution of Ca or Mn) and light-harvesting complexes of PS II as targets for Zn²⁺ toxicity,⁴⁸ conforming earlier results.⁴⁹ Decreases in non-cyclic photophosphorylation are probably due to inhibition of the electron transport chain rather than direct inhibition of ATP-synthase, because in this complex ion replacement is not likely.⁴⁹ Coherently, no significant effects on photophosphorylation were observed in Salvinia after Zn-rich wastewater exposure.⁴⁷ Reduction of the carboxylase capacity of RuBisCo suggests substitution of Mg²⁺ in the active centre of the enzyme.^50–52

As Zn²⁺ is an essential ion for normal plant growth, its uptake, storage and use is tightly regulated.⁵³ Under Zn²⁺ stress conditions (deficiency or toxicity), plants can regulate the expression of the relevant transporters, both at the transcriptional and post-transcriptional level.⁵⁴ Zinc (and cadmium) uptake over the plasma membrane is probably mediated via the ZIP transporter family (ZRT-IRT-like protein; zinc-regulated transporter and iron-regulated transporter protein). But uptake is possible as well via other transporters with similar affinities for different ions, e.g. iron transporters. A seemingly vicious circle develops: under iron limitation, plants up-regulate Fe uptake transporters, which then translocate more Zn, increasing the Zn content in the plant.^53,55 On the other hand, zinc deficient plants often accumulate more iron, proving the competition for uptake.⁵⁶ Similarly, a reduction in the content of other essential ions (such as Mg, Mn and Fe)⁵⁷ is recognized and enhances the expression of more such transporters. Thereby, the stress of deficiency in other ions enhances the stress of Zn²⁺ toxicity.

The homeostasis of Zn²⁺ is tightly regulated (see above) and it can be remobilized from storage in cases of limitation. Using “control” treatments without added Zn²⁺ will lead to Zn²⁺ deficiency unless chemicals of normal purity (p.a. grade) are used, which may contain enough Zn contamination for the specific species.^58,59

Although Zn²⁺ is present in the Cu/Zn SOD and thereby involved in the defence against oxidative stress, excess Zn²⁺ leads to the formation of reactive oxygen species. Both Zn²⁺ deficiency and toxicity enhanced hydrogen peroxide concentrations and SOD activity in mulberry leaves.⁵⁶ An increased ratio of dehydroascorbate to ascorbate (DHA-to-AsA) indicated a disturbed redox-status, shifting towards more oxidized forms in mulberry⁵⁶ and rapeseed seedlings.⁵⁷ The induction of ROS by biologically redox-inert zinc is not surprising in view of the many sites in photosynthetic light reactions where reactive oxygen species can be produced.⁶⁰ Also, the likelihood of such ROS production increases if the normal pathway of excitons and electrons in photosynthesis is inhibited. Even under optimal conditions, ROS arise constantly in metabolism pathways involving oxygen⁶¹ and act as messenger molecules.^55,62 Furthermore, under suboptimal conditions, ROS scavenging becomes inhibited. This can result from the substitution of Mg²⁺ in Chl (see above) because [Zn]-Chl has a reduced efficiency with respect to quenching of singlet oxygen.⁶³ Additionally, disturbances in other essential nutrients (Cu, Fe and Mg) by excess Zn²⁺ can lead to less functional or reduced activity of Cu/Zn-SOD, Mn-SOD and Fe-SOD.⁵⁷ However, Zn²⁺ concentrations in this study were far beyond environmentally relevant conditions (up to 1.12 mM), while hardly any effects were observed at the lowest concentration of 70 μM. This suggests either too short an exposure period to observe effects at environmentally relevant concentrations or a high chelation capacity in the nutrient solution, leading to a far lower bioavailability of zinc. A higher capacity and activity of the antioxidant system in zinc-tolerant hyperaccumulating ecotypes of Sedum alfredii over the non-accumulating ecotype suggests an involvement of the antioxidant system in mediating tolerance.⁶⁴ Exogenous ethephon, a precursor of the phytohormone ethylene, reversed negative effects of Zn²⁺ and Ni²⁺ (200 mg kg⁻¹ soil) on mustard plants through induction of the antioxidant system. Activities of SOD, APX and GR were increased due to Zn²⁺ or Ni²⁺ treatment compared to the control but were even higher after ethylene treatment and highest after Zn²⁺ or Ni²⁺ plus ethylene treatment.⁶⁵

Is cadmium only toxic?

For many decades, cadmium (Cd) has been known to be a highly toxic metal, not only to plants but also to animals, incl. humans. Concentrations in the soil do not usually exceed 5 mg kg⁻¹ in urban soils and are at low nM concentrations in aquatic environments (Table 2). Even in a heavily contaminated stream in Nigeria, the highest Cd concentration was 1.4 μM, clearly exceeding the limits set by the Federal Environmental Protection Agency.⁶⁶ However, this emphasizes that high μM or even mM concentrations are very rare.

Cadmium toxicity mainly originates from non-functional binding to biological ligands that are meant to bind other divalent metals. These are most often the amino acids cysteine (Cys) and histidine (His) as the most common amino acid residues in various metal centres of enzymes, in particular Zn centres due to the chemical similarity of the metals^67,68 (see below) but also RuBisCo as an enzyme with Mg²⁺ in its catalytic centre. Such Cd-substituted enzymes are usually non-functional. It is less known that such a ligand can also be chlorophyll, where Cd²⁺ easily replaces Mg²⁺ as the central ion.⁶⁹ In this case, damage to the affected organism originates from degradation of this pigment, which bleaches easily, but also from the unsuitability of Cd²⁺ for photosynthesis. As explained already for [Zn]-Chl, [Cd]-Chl also binds axial ligands with a much lower affinity than [Mg]-Chl, leading to protein denaturation. Furthermore, [Cd]-Chl quickly dissipates almost all absorbed excitation energy as heat due to its highly unstable singlet excited state (its lifetime is even shorter than that of [Zn]-Chl).³⁶ Because of the instability of [Cd]-Chl, leading to degradation during extraction/separation, as well as the very high similarity of its UV/Vis absorption spectrum with that of [Mg]-Chl, it is very difficult to actually measure [Cd]-Chl formation in plants.⁶⁹ Details of Mg-substitution were reviewed by Küpper et al. (2006).⁷⁰ Very recently, Cd incorporation into the major light harvesting complex, LHC II, has been shown to occur from 5 nM onwards in the shoot of the aquatic model plant Ceratophyllum demersum.⁷¹ In the absence of other potential high-affinity binding sites in this protein, this is most likely due to formation of [Cd]-Chl. Chlorophyll fluorescence data, especially F_v/F_m, were less affected by Cd exposure under the tested low light conditions, further suggesting incorporation into the LHCs.⁷⁰ Under high light conditions, the insertion of Cd into the PS II RC (presumably into the pheophytin)⁷⁰ is more likely, leading to the loss of the whole photosystem and thereby causing a prominent decrease in F_v/F_m, which was observed under high-light but not under low-light conditions.⁷¹

Analysis of Cd²⁺-induced chronic inhibition of photosynthesis in Noccaea (formerly Thlaspi) caerulescens indicated that Cd²⁺ inhibits photosynthetic light reactions more severely than the Calvin–Benson cycle.⁷² Furthermore, spectrally resolved analysis of photochemical vs. non-photochemical quenching in the same study showed that Cd inhibits at least two different targets in or around PS II.

Because of their chemical similarity, many transporters for divalent ions like Zn²⁺,^53,73 but also Ca²⁺ channels,⁷⁴ facilitate Cd²⁺ uptake into the roots and further distribution in the plants. The competition for binding sites can reduce the uptake of essential ions (such as Cu, Fe, Mg, Mn and Zn) into the roots and cause deficiency or even dislodge bound Zn²⁺ from binding sites and thereby change the tightly regulated zinc homeostasis in plant cells.⁵³ Lethally toxic Cd²⁺ (200 nM) concentrations did not affect the total accumulation of Zn²⁺ in the tissue of C. demersum but did affect its distribution: Cd²⁺ apparently inhibited the export of Zn²⁺ out of the vein, leading to Zn²⁺ deficiency in the mesophyll and Zn²⁺ toxicity in the veins of plants exposed to Cd²⁺.⁷⁵

A mechanistic uptake study of radiolabelled Cd or Zn in bread wheat (low Cd accumulations) and durum wheat (increased Cd accumulations) root cells revealed mutual uptake inhibition of both ions at the root cell membranes.⁷⁶ At the tested concentrations (50 nM–1.5 μM for Cd and 50 nM–50 μM for Zn), both metals yielded non-saturating uptake curves, with higher K_m values for the non-essential Cd, possibly because high-affinity transporters exist that are specific for Zn.⁷⁷

Another approach identified endogenously produced NO in root cells of A. thaliana as an important signalling molecule under Cd exposure, mediating Cd stress.⁷⁸ Enhanced fluorescence caused by an NO-sensitive dye was detected in roots and shoots (leaves and leaf discs). Using specific mutants, the authors could exclude the involvement of AtNOA1 and NR as catalysts for NO production. With additional microarray studies, they found a number of NO-dependent genes that were differentially expressed due to Cd exposure. Among the up-regulated genes they found IRT1. This gene encodes an iron transporter in the plasma membrane, which also transports Cd. Apparently, a cellular pathway resembling that of iron deprivation is mediated by NO, giving rise to Cd toxicity.⁷⁸

Although not redox active, Cd²⁺ exposure leads to enhanced production of reactive oxygen species.^79–81 One likely reason is the enhanced mis-transfer of electrons on oxygen instead of their target molecule, e.g. by [Cd]-Chl (see above). Another possible reason is that Cd²⁺ exposure reduces the capability of scavenging ROS. Several enzymes and non-enzymatic antioxidants are present in plant cells⁸² but Cd²⁺ treatment can alter synthesis or activity, leading to oxidative stress, both in roots and shoots. The replacement of Zn²⁺ in the Cu/Zn-SOD by Cd²⁺ changes the structure of the enzyme,^80,83 making it functionless and leading to its degradation (Cu^+/2+ is the redox-active ion but Zn²⁺ is believed to have structural purposes). Upon Cd²⁺ exposure, decreased contents and/or activities of the SODs were found.^80,84,85 However, depending on the Cd²⁺ concentration, increased activities of enzymes or numbers of isoenzymes were observed, indicating the protective role of the antioxidant system against moderate Cd²⁺ stress.^86–89 Further information on metal and specifically Cd²⁺-induced oxidative stress in plants and algae can be found in Sandalio et al. (2009)⁷⁹ and Pinto et al. (2003).⁹⁰ If the ROS are not detoxified in time, they can cause oxidation of membranes (lipid peroxidation) and produce mutagenic aldehydes.⁹¹ Furthermore, the direct interaction of Cd²⁺ (and other metal ions) with the nucleotides^92,93 or the inhibition of DNA-repairing enzymes can induce DNA damage.⁹⁴ An A. thaliana mutation assay revealed an increasing number of point mutations from very low Cd²⁺ concentrations (8.8 nM), while, e.g. Cu²⁺ and Ni²⁺, had much less potential to induce the mutations.⁹⁵ Many genotoxicity assays (for plants) are designed to test contaminated soils or waters and the applied metal concentrations are often beyond natural conditions. The order of metals inducing mutagenic effects was Hg²⁺, Cd²⁺ (10⁻⁷–10⁻⁵ M) > Zn²⁺, Pb²⁺, Cu²⁺, Ni²⁺, Co²⁺, Al³⁺, Cr³⁺ (10⁻⁴–10⁻³ M) > Mn²⁺, Mg²⁺ (10⁻² M) based on the occurrence of micronuclei in onion root tips.⁹⁶ Later, other systems were found to be more sensitive, (Tradescantia < Vicia faba < transgenic A. thaliana).^95,97 Therefore, it is not always easy to tell how much genotoxicity adds to phytotoxicity under environmentally relevant conditions.

Plants are immobile and cannot avoid unfavourable heavy metal concentrations in soils. They have developed several ways of detoxification, including chelation, immobilization, exclusion and compartmentalization.^98,99 One major group of ligands are the enzymatically synthesized phytochelatins (PCs),^100,101 which are induced most efficiently by the nonessential metal(loid)s Cd and As but also by Ag, Pb, Cu, Hg, Zn, Sn and Au.^73,100,101 PC–metal-complexes are probably transported into the vacuole, where the metals cannot interfere with the photosynthetic and respiratory complexes.⁷³

Sequestration into the vacuole is known for many more metals in hyperaccumulating and non-hyperaccumulating plants, although in hyperaccumulators, metals are bound to weak ligands, not phytochelatins (see Leitenmaier & Küpper, 2013 for a recent review).⁹ In some plants, special storage cells are located in the epidermis and the metals need to be transported from roots to tissues above ground level, requiring many translocation steps against a concentration gradient. Metal concentrations of up to hundreds of mM in the vacuole were reported for hyperaccumulating plants.^102,103 The respective transporters have been partially characterized as summarized in Clemens (2006)⁷³ and Leitenmaier & Küpper (2013).⁹ When the rootless macrophyte, Ceratophyllum demersum, was exposed to very low Cd concentrations (2 nM) for 3 weeks, a homogenous distribution of Cd over the whole cross section of the leaf area was found and phytochelatins had not been induced. After prolonged exposure (6 weeks) increased sequestration of Cd into the epidermis was found. This indicates the onset of detoxification by sequestration even at very low chronic toxicity.⁷⁵

Expression analyses revealed up- or down-regulation of various genes in response to Cd exposure.¹⁰⁴ Although the reason for this is not always clear and it is not necessarily directly caused by Cd-toxicity (one should be careful with high Cd concentrations), one can draw conclusions on how their regulation may be affected by Cd toxicity. As an example, some transcription factors that are up-regulated in response to Cd in Arabidopsis thaliana are constitutively strongly expressed in the Cd-hyperaccumulator plant Arabidopsis halleri.¹⁰⁵ Furthermore, miRNAs, small RNAs that are usually involved in negative gene regulation by destroying (“silencing”) their respective mRNAs, can act as stress regulators after Cd exposure.¹⁰⁶ The miRNA miR393 targets E3 ubiquitin ligase/TIR1 (transport inhibitor response1), leading to less mRNA and thereby down-regulation of auxin signalling and possibly less proteolysis of the respective ubiquitin targets.¹⁰⁶

To avoid patches of unfavourably elevated Cd content in the soil, roots can adapt by enhanced lignification and production of suberin lamellae at the sides facing the Cd contamination.¹⁰⁷ The authors state that these local barriers could restrict the apoplasmic movement of Cd and thereby the Cd loading into the xylem and its further transport into other root and shoot tissues.

In more recent times, however, at least for some organisms, a beneficial role of cadmium could be convincingly shown. This is, as the most prominent case, the expression of an alternative isoform of carbonic anhydrase, which in contrast to the regular isoform works well with Cd²⁺ instead of Zn²⁺ in its active centre. Originally it was found in the marine alga Thalassiosira weissflogii,¹⁰⁸ from where it was purified, spectroscopically characterised and ultimately crystallised.^109–111 It was later found in other algae as well,¹¹¹ showing that it probably evolved as a remedy against the widespread extreme zinc limitation in the oceans and that in rare cases, Zn²⁺ can be functionally substituted by Cd²⁺.¹¹¹ Finally, it may even occur in Cd-hypertolerant terrestrial plants but in this case probably for a different reason, more comparable to the occurrence of [Zn]-BChl instead of Mg-BChl in Acidiphilium (see above). In N. caerulescens, the use of Cd-carboanhydrase could prevent uncontrolled exchange of Zn²⁺ by Cd²⁺ in normally Zn-containing carboanhydrase. This would explain why, in this species, Cd²⁺ induced carboanhydrase activity, while in a related non-hyperaccumulator species, Cd²⁺ decreased it.¹¹² The positive effect of Cd²⁺ here can be traced back to the Cd²⁺-containing enzyme. For many other metals (and other chemicals), positive effects have been found when they were applied in minute concentrations. This is usually attributed to the hormetic effect, which represents an overcompensation response in the treated organism, thereby triggering favourable effects instead of toxicity.¹¹³

Mercury – one of the most toxic metal ions

For mercury (Hg), so far no beneficial biological role has been found in any organism, while it is known to be among the most toxic metal ions for all organisms. Zones that are highly contaminated due to natural Hg sources can be found in Europe (the Almadén district in Spain and the mercury mines in Idrija, Slovenija) and in China (Gouxi in the Guizhou Province) with concentrations up to 76 μg g⁻¹ in Slovenija¹¹⁴ and 2 μg g⁻¹ in China.¹¹⁵ For comparison, the NYS DEC set a maximum of 0.81 μg g⁻¹ for agricultural soils (Table 1). Even if concentrations of bioavailable Hg are lower, the contaminated areas pose a health risk for humans: the surroundings of the Chinese mining districts are used for rice production.^116,117 The methylmercury (MeHg) accumulated in the rice seeds acts as a neurotoxin.¹¹⁸

Soluble mercury in the environment and organisms occurs in almost all cases as Hg²⁺; only some bacteria have an enzyme that is able to reduce it (to Hg⁰, metallic mercury), which was also used for making transgenic plants with this property.^119,120 Mercury reduction was also observed in many phytoplankton species, but how they reduce it is still unknown.¹²¹ Natural mercury concentrations in the oceans rank between 1 and 100 pM.^121,122 Soils other than those in polluted areas range between 20–150 ppb but fertilizers and manures contaminated with Hg (and other toxic metals) can increase concentrations drastically.¹²³

The toxicity of Hg²⁺ is to a large extent caused by its chemical similarity to zinc (Zn), which it can replace in active sites, especially those with imidazole N and thiolate S ligands.⁶⁷ This similarity furthermore facilitates the uptake of mercury ions into roots or algal cells via transporters for other essential ions^124,125 and leads to the replacement of other divalent metal ions in their active sites, including Mg²⁺ in Chl.⁶⁹ Many studies showed a decrease in Chl content due to Hg²⁺ exposure,^126–128 although initial increases can occur as well. In the early stages of Hg²⁺ exposure (14 days, 100, 200 and 500 mg kg⁻¹ soil, values for contaminated sites are usually less than 100 mg kg⁻¹, Table 1), the Chl content was increased in winter wheat compared to untreated control plants.¹²⁹ With longer exposure time (28–34 days), Chl was reduced in all samples exposed to Hg²⁺. However, plant species, experimental conditions and importantly, Hg concentrations used matter. P. glomerata plantlets exposed to only 1 μM of Hg²⁺ had Chl contents similar to the control but showed an increased activity of the enzyme δ-ALA-d (delta-aminolaevulinic acid dehydratase), which is involved in the biosynthesis of tetrapyrroles, probably balancing the degradation of Chl with enhanced biosynthesis.¹³⁰ Earlier works showed that this enzyme can be inhibited by Hg, Pb, Cd and Zn.^131,132 The enzyme NADPH:protochlorophyllide oxidase (POR), which performs photoreduction of protochlorophyllide into chloropyllide, is inhibited by Hg²⁺.¹³³ However, leaf homogenates were incubated with very high Hg²⁺ concentrations (there was no effect visible below 10⁻³ M due to very short incubation times of max. 3 h) and the authors doubt that Hg²⁺ ions would react with the enzyme in intact plants.¹³³ Conclusions about toxicity mechanisms should therefore be taken with care.¹³⁴ Similarly, most aquaporins are blocked by Hg²⁺ ions binding to the sulfhydryl group of Cys residues close to the aqueous pore, reducing the hydraulic permeability of root cells (detailed review by Javot and Maurel, 2002).¹³⁵ Binding to nitrogen in the imidazole ring of His was shown as well.¹³⁶ However, how far the blockage of aquaporin contributes to mercury toxicity under environmentally relevant conditions remains unclear, because many experiments were performed as studies to characterize the aquaporins, not as studies of Hg²⁺ toxicity (e.g. expression in Xenopus oocytes or other cell types).^135,137

Mangroves grown in Hg-amended soil for 12 months showed significantly reduced F_v/F_m values only at the highest Hg²⁺ concentration (160 μg g⁻¹), although the Chl content was decreased from 40 μg g⁻¹ onwards and hardly any Hg was translocated from the roots into the leaves.¹²⁸ However, changes in both F₀ and F_m (F_v = F_m − F₀) can lead to unchanged F_v/F_m for complex reasons.¹³⁸ For example, photosystems that became non-fluorescent because of the formation of [Hg]-Chl do not contribute to this parameter although they would be non-functional like other heavy-metal-substituted chlorophylls.

Photosynthetic oxygen evolution and CO₂ fixation (determined with ¹⁴C) declined with increasing mercury concentration (0.5–3 μM) in Nostoc muscorum, while respiration increased dramatically compared to control conditions.¹³⁹ PS II was more affected than PS I, as shown for many other metals (reviewed e.g. by Küpper and Kroneck, 2005).⁸ Tests with exogenous electron donors indicated the inhibition site to be between the oxygen evolving complex (OEC) and PS II. The negative effects were more pronounced under high light conditions,¹³⁹ which is in line with previously reported substitution of Mg in reaction centre Chl.⁶⁹ A trend of increasing lifetime of the Chl autofluorescence was observed in diatoms exposed to Hg²⁺, while no change occurred after exposure to MeHg.¹⁴⁰ This emphasizes the different toxicity mechanisms of organic vs. inorganic mercury. Organic mercury cannot substitute other ions in photosynthetic complexes. However, the results were not tested for statistical differences, which unfortunately is a problem in many studies. Spikes in both directions can be easily misinterpreted as true signals. The consequences of Hg²⁺ insertion in Chl and in proteins are very similar to those triggered by other heavy metals. Mercury-treated plants have higher amounts of reactive oxygen species (ROS) and molecules and enzymes of their antioxidant system components, resulting from the mis-transfer of electrons on oxygen instead of their electron acceptor.⁸² Cultures of C. reinhardtii that were exposed to 1–6 μM of HgCl₂ for 24 h showed progressively increasing activities of SOD, APX and CAT, while only the highest concentration (8 μM) led to activities lower than that of the control treatment.¹²⁷ Not only did the activities change due to Hg-treatment, but the expression levels of the respective genes coding for Mn-SOD, APX and CAT were upregulated as well. Comparable results were found for plantlets of Pfaffia glomerata shoots, while enzyme activities in the roots of P. glomerata and rice decreased with increasing Hg concentrations.^130,141 Depending on the applied Hg concentration and treatment duration, changes in this pattern are possible. Generally, the induction of ROS (hydroxyl anion OH˙, superoxide anion O₂˙⁻ and hydrogen peroxide H₂O₂), followed by lipid peroxidation and decreased membrane integrity seem to occur after Hg treatment.^126,142,143 However, at very high Hg concentrations (and these may be specific for certain plants), an overall inhibition occurs, which is no longer specific to Hg toxicity. Even so, many mercury compounds can induce genotoxicity in plants, including chromosomal aberrations, polyploidy and the occurrence of micronuclei.¹⁴⁴ However, as described for Cd²⁺, these tests were performed as risk assessment studies, not to unravel mercury toxicity mechanisms.

Non-enzymatic antioxidants, proline and especially thiol compounds are induced by Hg stress. Dago et al. (2014) extracted glutathione and phytochelatins (PCs) from wild asparagus from the ancient mercury mines in Spain.¹⁴⁵ Higher phytoavailable Hg²⁺ concentrations were correlated to higher concentrations of the PCs.¹⁴⁵ Longer-chain PCs were found in the roots and shorter ones, especially PC3, in the aerial parts of the plants. Roots generally possessed higher concentrations. As the plant material was ground, however, Hg–PC complexes could have been formed when the vacuole was disrupted. Thus, from such studies it cannot be determined for sure whether the Hg–PC complexes were formed inside the plants but only that Hg²⁺ induced the synthesis of PCs that could be potential ligands for Hg²⁺.

Generally, accumulation of mercury seems to be higher in the roots than in the shoots of exposed plants.^{123,125,126,146} This could be either a successful translocation stop with the roots acting as a barrier towards the toxic metal or the blockage of metal transporters, which would lead to more stress as transport of essential metals into the above-ground tissues would be inhibited. To reveal the fate of mercury ions in the roots, X-ray fluorescence (XRF)-related techniques allow identification of the tissues or organs in which the mercury binds preferentially. Carrasco-Gil et al. (2013) used μ-XRF to map the distribution and extended X-ray absorption fine structure (EXAFS) analysis to determine the speciation of mercury in roots of Medicago sativa and Marrubium vulgare.¹⁴⁷ While the first sample was hydroponically grown and exposed to Hg²⁺ under controlled conditions, the latter was collected from a mercury-contaminated area in Spain. The distribution of Hg was different in both species. In M. sativa, Hg was found in the apical regions of primary and secondary roots, in the vascular cylinder and the epidermis. EXAFS spectra revealed that a high proportion of mercury was bound to organic thiols like phytochelatins and GSH.¹⁴⁷ The plant from the contaminated site had a different distribution and speciation of mercury in the tissues. The detected HgS minerals may actually have been from soil microparticles still sticking to the roots.¹⁴⁷ A slightly different setup was used by Debeljak et al.: roots of maize plants grown in Hg-amended soils (50 mg kg⁻¹) were dipped in a special medium that does not penetrate the tissues, rapidly frozen, sectioned with a cryotome and then freeze-dried.¹⁴⁸ The sections were analysed with laser-ablation inductively coupled plasma mass spectroscopy (LA-ICPMS). The highest Hg concentrations were found in the outer part of the roots, the epidermis and the endodermis, suggesting that Hg ions cannot cross the endodermal barrier and be transported to the upper part of the plant,¹⁴⁸ which is consistent with many other studies. Usually, more Hg was found in the roots than the shoots and leaves of the plants (see above). However, both studies only allow for the detection of ions on the tissue, not at the cellular level, because samples were freeze-dried. Upon freeze-drying, the vacuole becomes air-filled. Since it does not have an internal solid matrix that could keep them in the middle of the vacuole, solutes that were in the vacuole will stick to the tonoplast membrane surrounding the vacuole. In a dried plant cell, where the cytoplasmic layer is extremely thin, at resolutions achievable with current metal analysis techniques this will be indistinguishable from binding to the cell wall. Possible occurrence of such artefacts should always be checked by measuring an abundant metal with well-known intracellular accumulation (e.g. potassium) as a natural internal reference. True specific cell-wall binding of Al was proven for frozen-hydrated tea leaves.¹⁴⁹ Sub-cellular fractionation of tissues also poses the risk of artefacts. The step of homogenization of tissues and breaking open of cells brings all cell components into close contact. The results obtained from this technique are potential binding sites for heavy metals, which are not necessarily the occupied binding sites in the intact plant tissues.¹²⁵ For example, plant cell walls are composed of compounds that can bind divalent and trivalent ions very effectively. Nevertheless, in many cases where binding of the majority of the metal to cell walls was actually a result of sample preparation artefacts, cell walls certainly still play a role in binding, uptake, transport and detoxification of trace metals – they are just not the major storage site.¹⁵⁰

Recently, the effect of Hg on plants and algae at the level of microRNAs, genome-wide transcriptomics and signalling molecules (NO, CO and salicylic acid) has been reviewed by Chen and Yang (2012).¹³⁴ Briefly, mercury exposure triggered the expression or up-regulation of the general biological defence system, chlorophyll synthesis, cell wall metabolism, biosynthesis of secondary metabolites and Hg tolerance.

Toxicity of biologically redox-active metal ions

Copper – among the most needed but also among the most toxic metals for plants

Copper is among those “heavy metals” that have been known for a long time to be essential micronutrients, while also easily reaching toxic levels. In contrast to animals incl. humans, for plants copper is even more toxic than cadmium, as shown by many studies where both metals were compared.^69,151 Even in the open ocean, where organisms rarely suffer from toxicity but frequently from deficiency of micronutrients, copper reaches toxic levels, for example in the Sargasso sea, where its natural abundance limits growth of cyanobacteria.¹⁵² In freshwater ecosystems, copper toxicity most often occurs as a result of human activities, which fall into two groups – industry and agriculture. Industrial contamination from various individual sources led to toxic copper levels in major rivers in Western Europe, e.g. the Rhine with concentrations of up to 500 nM in the 1970s. Such concentrations are lethal to sensitive species of cyanobacteria and plants.^5,153 This type of contamination has drastically decreased due to better industrial practices and wastewater treatment. The second source of severe copper contamination, in aquatic ecosystems as well as in soils, however, remains (see Table 1). This is the use of copper-containing pesticides in agriculture, in particular in vineyards. This can lead to very high pollution levels with hundreds of ppm of Cu in the soil.^4,154–156

Copper toxicity in photosynthetic organisms has been investigated for several decades, leading to a detailed understanding as reviewed ten years ago (Küpper and Kroneck, 2005).⁸ However, more recent research yielded further significant new insights and there are still important open questions. As for the other metals discussed in this review, mechanisms of copper toxicity have often been studied using extremely high, environmentally irrelevant concentrations. This applies in particular to many older articles, but those that are frequently cited include Gallego et al. (1996)¹⁵⁷ and Wecks and Clijsters (1996),¹⁵⁸ which are often cited in relation to oxidative stress caused by copper toxicity. Studies that used much lower but still toxic copper concentrations came to completely different conclusions concerning the main mechanism of copper toxicity in plants. As a prime target, in many studies with low copper concentrations, photosynthetic light reactions were found. Inside the photosynthetic system, several targets were identified. Generally, PS II was found to be much more sensitive than PS I. In PS II, under high irradiance (but not related to photoinhibition), the reaction centre was found to be the prime target, while under low irradiance, copper caused malfunctioning of LHC II by substitution of Mg²⁺ by Cu²⁺ in its chlorophyll.^69,159,160 As already described for Cd²⁺, Hg²⁺ and Zn²⁺, this leads to enhanced thermal dissipation of captured excitons, because like the aforementioned metals, [Cu]-Chl has an unstable singlet excited state. However, for [Cu]-Chl it is even less stable, so that all captured excitons are thermally relaxed.³⁶ The exact target site of copper toxicity inside the PS II reaction centre has been a topic of intense research. Inhibition of primary charge separation in the PS II reaction centre was first suggested by Hsu and Lee (1988)¹⁶¹ and would make sense in terms of insertion of copper into the pheophytin.¹⁵⁹ A series of detailed mechanistic studies on Cu toxicity with respect to the PS II reaction centre was also performed by Yruela et al. in the first half of the 1990s, representing the state of knowledge for various individual aspects of this toxicity.^162–165 Using spectroscopic methods, these authors clearly show that under their experimental conditions, copper binds in the pheophytin a – Q_a domain and that copper competes with protons for binding. However, they worked on isolated thylakoids, at relatively high copper concentrations (5–100 μM) and with reduction by dithionite, leading to uncertainty regarding how far the reaction would reproduce that in a living plant. Copper is one of the classical redox active trace elements in biological systems, as such being an essential cofactor for many enzymes. Therefore, it was to be expected that copper toxicity enhances ROS production. This has been reported in many studies, as mentioned earlier;⁸ thus, this review will focus on more recent insights. In most of the earlier studies, lethal copper concentrations were applied and it remained unclear whether the ROS production was a cause or consequence of the inhibition of activity, e.g. of photosynthesis. Recently, copper toxicity was re-investigated using sub-lethal low nanomolar concentrations causing chronic toxicity in the shoot of the aquatic model plant, Ceratophyllum demersum, under conditions simulating oligotrophic lakes.⁵ In this study, copper toxicity first of all affected the PS II reaction centre (see above), while ROS production seemed to be a consequence as it occurred later. In contrast, in an earlier study on copper stress in green algae, 50 and 250 nM copper led to ROS formation, which then led to inhibition of photosynthesis as demonstrated by restoring photosynthesis via an ROS scavenger.¹⁶⁶ In other recent studies on Matricaria chamomilla and Arabidopsis thaliana using micromolar concentrations that caused acute toxicity, ROS production occurred within a few hours in the roots, i.e. organs without photosynthesis.^151,167 In a study comparing heterotrophic (white) with photosynthetic cells of the same strain of the alga Euglena gracilis, ROS production in response to Cu and Cr toxicity was much higher in the photosynthetic cells.¹⁶⁸ In summary, it seems that a basic level of ROS production in response to Cu (and Cr) toxicity is reached directly, without photosynthesis. Damage to photosynthesis strongly increases ROS production, while directly copper-induced ROS production in turn inhibits photosynthesis.

Besides the intensively investigated inhibition of photosynthesis, according to several studies on different species, copper toxicity also disturbs nutrient uptake. While differences in these disturbances exist even among ecotypes of the same species,¹⁶⁹ in all cases copper toxicity caused a decrease in iron content in the shoots,^169,170 while acclimation to copper toxicity included a recovery of iron concentrations.^170,171 Nanomolar chronically toxic Cu²⁺ was furthermore observed to inhibit zinc uptake.⁵ Comparison of ecotypes suggests that some of these changes in the shoot are caused by changed uptake in the root.¹⁶⁹ A study by Thomas et al. (2013) was performed on the rootless submerged shoots of the model plant Ceratophyllum demersum, clearly showing that nutrient uptake/distribution in the shoot is affected as well.⁵ In roots of rice, it was recently shown that copper interacts with vesicle transport.¹⁷² By knockout of genes necessary for this vesicle transport, the authors furthermore found that this vesicle transport is essential for signalling via ROS for activating defences.¹⁷² Further, signalling via nitric oxide (NO) seems to occur during copper toxicity; this was shown to induce proline synthesis at low micromolar copper concentrations, inhibiting growth of Chlamydomonas.¹⁷³ Proline synthesis has been known for a long time to be a defence reaction, not only against copper toxicity but also other stresses.¹⁷⁴

In terrestrial plants, roots are the first organs to come into contact with an excess of Cu and often they accumulate much higher copper concentrations than the shoots, so that they become a primary target for damage.^175,176 This usually results in a decrease in biomass.¹⁷⁷ Changes in the root morphology, numbers of root hairs¹⁷⁸ or cell volume¹⁷⁹ are signs of Cu toxicity. A decreased number of root tips or organelles (like mitochondria) within root cells¹⁸⁰ indicates stress and will lead to a generally decreased energy production, starch accumulation and finally biomass. However, these findings are consequences of various Cu toxicity mechanisms and do not per se represent mechanisms of Cu toxicity to roots. A study by Pätsikkä et al. (2002)¹⁸¹ showed that competition with iron uptake is one of the mechanisms of Cu-induced damage in roots.¹⁸¹ Some more details on how Cu influences these changes in root morphology (root system architecture) and growth of primary and lateral roots were done in Arabidopsis thaliana.¹⁷¹ Using fusion constructs of specific growth markers with the reporter gene GUS, the authors showed reduced mitotic activity in the respective root tips under Cu stress. The involvement of phytohormone accumulation in inducing (auxin) or inhibiting (cytokinin) lateral root growth at different Cu concentrations was shown, but how Cu induces this (e.g. binding sites of Cu and gene transcription) is not yet known.¹⁷¹ Defence against copper toxicity on roots involves efflux pumps.¹⁸² Diminishing passive copper inflow by enhanced root lignification, mediated via up-regulation of peroxidase expression, probably plays a role as well.¹⁸³

While all research on copper toxicity mentioned so far had been carried out with dissolved copper, in recent years studies on copper (usually CuO) nanoparticles were undertaken. However, in most cases it remained unclear how far the nanoparticles dissolved during the experiment and whether the plants actually took up any nanoparticles or only dissolved copper. Even in the rare cases where dissolution of copper from nanoparticles was measured,¹⁸⁴ it was not done under conditions relevant for soil, nutrient solutions or inside plants, rendering these measurements useless. Also, typical treatment concentrations of 10 000 to 1 000 000 ppb CuO nanoparticles¹⁸⁴ are very high compared to the roughly 1 ppb at which copper toxicity may start in sensitive organisms.⁵ Thus, the relevance of significant DNA damage, specifically by copper nanoparticles, needs to be re-investigated.¹⁸⁴ In another, more recent example of such a study, termed “mechanistic” by the authors, the effects of the CuO nanoparticles matched known effects of dissolved copper, such as general growth inhibition, pigment loss, ROS production and root lignification.¹⁸⁵ A few years before, however, it had been found that polymer-coated CuO nanoparticles are more toxic to algae than the same particles without coating, because the coated particles could cross membranes more easily.¹⁸⁶ In another recent study, the authors were able to see some particle-specific effects and furthermore characterised the aggregation and dissolution of the nano CuO in the tested media.¹⁸⁷ The particle-specific effects happened in the very early parts of the response, within the first five hours. It thus remains to be seen how far these effects are environmentally relevant, since effects occur within such short treatment times only at rather high concentrations. In any case, this study was very informative concerning the behaviour of CuO nanoparticles in different plant growth media.

Iron – rarely but also severely, toxic

Iron toxicity is a topic not often dealt with in plant sciences (incl. algae and phototrophic bacteria), because in the oceans iron is always deficient and even in terrestrial plants, deficiency is more frequent than toxicity. This is due to the redox properties of iron – the abundant redox state in current atmospheric conditions on Earth is Fe(iii). This is hardly soluble and therefore mostly remains biologically inaccessible in minerals. The only chance for iron to become toxic is the reduction of massive amounts of Fe(iii) to Fe(ii), which makes it highly soluble. This phenomenon, however, does frequently occur in one very major crop species: rice. The soil of flooded lowland rice fields tends to become anoxic very quickly and it is often rich in iron. The same occurs in natural freshwater wetlands^188,189 and has recently been found for salt marshes as well, where iron toxicity to the halophyte Sueda maritima was described.¹⁹⁰ Once the soil becomes anoxic, the iron is reduced and bioavailable, as described in the review by Becker and Asch (2005).¹⁹¹ As written in that review, “iron toxicity remains an important constraint to rice production and, together with Zn deficiency, it is the most commonly observed micronutrient disorder in wetland rice”. While that review focussed on conditions and management of iron toxicity in rice, we now would like to focus on the current knowledge on mechanisms of iron toxicity in plants.

Iron toxicity was first described as a problem in rice 60 years ago.¹⁹² This publication accurately described the visible symptoms, such as brown spots on the leaves, and associated it with reducing conditions. The mechanisms behind the symptoms, however, remained unknown. Later it was shown that strong oxidative stress occurs in plants during iron toxicity.¹⁹³ During iron toxicity stress, various reactive oxygen species have been measured, such as hydroxyl radicals, superoxide radicals, singlet oxygen, hydrogen peroxide and alkoxyl and peroxyl radicals, as reviewed by Becana et al. (1998).¹⁹⁴ In the context of human physiology, it was postulated early on that iron toxicity originates from the generation of reactive oxygen species via the Fenton reaction or the iron-catalysed Haber–Weiss reaction.¹⁹⁵ However, these particular reactions were never really proven to occur in vivo, only the rise in reactive oxygen species has been measured. All publications on iron toxicity in plants have since relied on these postulated reactions.^193,194,196 Therefore, providing clear evidence of whether these or other reactions cause oxidative stress during iron toxicity in plants would be an important topic for future research. Furthermore, it still remains to be shown that the occurrence of reactive oxygen species during iron toxicity is actually the cause of inhibition and not a consequence of it. These points have to be mentioned in particular, since it turned out that in the case of other metal(loid)s, toxicity-induced oxidative stress was not directly caused by redox reactions of the metal but by the malfunctioning of metal-inhibited photosynthesis (see metals previously discussed in this review). A detailed early study on iron toxicity by Kampfenkel et al. (1995)¹⁹³ reports a decreased ratio of maximal to minimal chlorophyll fluorescence quantum yields, which at that time was interpreted as photoinhibitory damage to photosystem II. By now it is well known that this ratio (now usually published as F_v/F_m = (F_m − F₀)/F_m) is not a specific indicator of photoinhibition but generally shows the dark-adapted maximal quantum yield of PS II photochemistry, i.e. a decline of this ratio generally indicates damage to the PS II reaction centre. Such damage has been reported with respect to many types of metal toxicity (e.g. review by Küpper and Kroneck, 2005).⁸ The first study that showed the involvement of light in the generation of ROS during iron toxicity in plants originates from 1993.¹⁹⁷ These authors reported that susceptibility to photoinhibition was increased by iron toxicity and showed that iron toxicity symptoms were absent when the plants were grown in very low light. They postulated that non-heme iron (as present e.g. in the PS II reaction centre) was responsible for generation of ROS in chloroplasts (mainly via singlet oxygen). Direct evidence for a malfunction of PS II causing the generation of reactive oxygen species under iron toxicity was provided by Suh et al. (2002),¹⁹⁸ who showed that iron toxicity causes increased synthesis of Cyt b₆/f to an extent that it produces singlet oxygen via photodynamic action, leading to inhibition of PS II.

Iron uptake into the plants under iron toxicity conditions is different from the better known pathway under iron-deficient conditions, where plants developed strategies to enhance it by soil acidification and exudation of mugineic acid as a siderophore. The worst toxicity was recently reported for feeding the iron as Fe(ii) sulphate, although Fe(iii) citrate was taken up in larger quantities and transported more efficiently from the root to the shoot.¹⁹⁹ This may be due to a so far unknown interaction of ferrous iron with transport proteins for other nutrients and minerals. This thought is supported by earlier results showing that iron toxicity increases uptake of sodium and interacts in a more complicated way with the uptake of calcium, magnesium, manganese, molybdenum, phosphorus and zinc.²⁰⁰ Revealing the mechanistic details of these interactions will be an interesting topic for future research.

Defence against iron toxicity is known to involve active oxidation in the rhizosphere in order to produce insoluble Fe(iii) minerals.²⁰¹ This has recently been confirmed on the genetic level, where a main quantitative trait locus (QTL) for iron tolerance was modifying root architecture towards conducive air transport into the roots.²⁰² Furthermore, the precipitated iron was now described as “iron nanoparticles” by comparison with artificial iron nanoparticles.²⁰³ Once iron toxicity has started inside the shoots, plants up-regulate enzymes that detoxify reactive oxygen species.^193,194 The pool of weakly bound iron in plants is controlled by the iron-binding protein ferritin, so that its ectopic over-expression leads to enhanced resistance against iron toxicity,¹⁹⁶ and iron ferritin protein levels are up-regulated during iron toxicity stress.²⁰⁴ Transporters pump iron out of the sensitive cytoplasm into compartments where it does less harm, as shown by the increased resistance towards iron toxicity upon over-expression of AtNRAMP1.²⁰⁵ Expression of the iron transporter YSL-1 spreads from the xylem parenchyma to the mesophyll under iron toxicity stress, which was interpreted as re-distributing excess iron to cells with more potential to detoxify it e.g. via vacuolar sequestration.²⁰⁶ Interestingly, iron toxicity elicits strong ethylene signalling, which in a still unknown way is important for efficient defence of the plant against toxicity.²⁰⁷

Nickel – an ultra-micronutrient with low toxicity

In plants, nickel is known to be needed for only one enzyme, urease, as reviewed e.g. by Küpper and Kroneck (2007)²⁰⁸ and Chen et al. (2009).²⁰⁹ For this reason, it is required by most plants only in minute quantities (usually 0.05–10 μg g⁻¹ dw in plants²¹⁰), so that the study of nickel deficiency involves a lot of effort in lowering the Ni²⁺ uptake by the plants to a critical level.^211,212 Only nickel hyperaccumulator plants, which use its toxicity as a defence against pathogens and herbivores, require much higher levels of nickel for normal growth.²⁰⁸

The low requirement for nickel is not paralleled, however, by a low threshold for toxicity. On the contrary, nickel is far (more than 100 times) less toxic to plants than the much-needed copper and most other trace elements, as can easily be seen in a comparison of various potentially toxic metals,⁶⁹ and as reviewed previously.²⁰⁸ For this reason, in many cases when “nickel toxicity” in the environment is reported, it is in reality toxicity of copper, which often occurs together with nickel.^213–215 Recently, synergistic effects of toxicity were also reported for a combination of nickel and cadmium. Concentrations of both metals, which did not cause toxicity on their own, led to severe toxicity when they were combined.²¹⁶ The reason for this synergistic action is not clear and cannot be deduced from current knowledge about the mechanisms of toxicity of the individual metals involved.

Pure nickel toxicity causes several distinct effects, which have been reviewed by Küpper and Kroneck (2007),²⁰⁸ so the current review will focus on those that were proven to be important under environmentally relevant low concentrations of nickel. Roots were shown to be sensitive to nickel toxicity, with inhibition measured at 2.5 μM nickel.²¹⁷ The mechanism of this inhibition remained unclear, as did subtle morphological changes in the rhizodermis, which were observed at 1 μM Ni²⁺.²¹⁸ Another root-level effect of nickel toxicity, which was shown at low micromolar concentrations, is inhibition of the uptake of nutrients.^219–222 This is probably due to interaction of Ni²⁺ with transport proteins. The exact mechanism remains to be resolved. In shoots of the submerged aquatic macrophyte Elodea canadensis, low micromolar concentrations of nickel were found to induce sublethal oxidative stress in terms of lipid peroxidation.²²³ It remained unclear, however, whether this oxidative stress was primary, i.e. directly caused by the Ni²⁺, or a secondary consequence, e.g. of malfunctioning photosynthesis, which was severely inhibited under the same conditions. Also, the inhibition of photosynthesis by exchange of Mg²⁺ against Ni²⁺ inside the chlorophyll was resolved all the way to the molecular level some years earlier. It was first reported in vivo by Küpper et al. (1996),⁶⁹ long after [Ni]-Chl had been shown to thermally dissipate all absorbed photons due to a very unstable excited state, as in the case of [Cu]-Chl (see above). This physical property of [Ni]-Chl (like [Cu]-Chl) makes the affected light harvesting systems act as “black holes” for excitons. For Ni²⁺, this was shown in a very detailed and thorough study on isolated bacterial photosystems.²²⁴ In that study, about three percent exchange of the central Mg²⁺ ions of all chlorophylls in the photosystem against Ni²⁺ were sufficient for complete inhibition of photosynthesis. Besides the thermal relaxation of excitons, the lack of axial ligands in [Ni]-Chl²²⁵ makes this pigment unusable for photosynthesis, as these axial ligands are required for proper folding of the pigment–protein complexes.^37,226

Combinations of metals

Most metal-polluted areas have too-high concentrations of more than one metal, especially around mining areas. Generally, the interactions of combined threats are synergistic (total effect is greater than the sum of individual compounds), antagonistic (total effect is lower than the sum of individual compounds) or additive (total effect equals sum of individual compounds). When exposed to binary mixtures of Cd, Cu and Pb from 40 to 640 mg kg⁻¹ each, Cucumis sativus exhibited all three responses (shoots: Cu + Cd and Cu + Pb: antagonistic, Cd + Pb additive; roots: Cu + Cd and Cu + Pb additive, Cd + Pb synergistic). In a tertiary mixture, however, only antagonistic responses were found.²²⁷ This determination was purely based on root and shoot growth, no physiological parameters were assessed.

Ince et al. (1999)²²⁸ used a statistical approach to predict interactions and found 87% antagonistic and 13% additive results for duckweed (Lemna minor) for various binary mixtures of metals (Co, Cr, Cu and Zn). A microtox assay (bacteria, Aliivibrio fischeri) yielded 41% antagonistic, 38% additive and 11% synergistic predictions.²²⁸ When exceeding optimal concentrations, additive or antagonistic effects seem to be the main responses. Again, the determination parameter was based on biomass and rather serves as a criterion for contamination determination, not to unravel toxicity mechanisms.

However, for macrophytes, a synergistic interaction based on various photosynthetic parameters was found for low concentrations of Ni (300 nM) and Cd (3 nM). While Cd only had positive effects and Ni was only slightly inhibitory, Cd and Ni together resulted in increased inhibition.²¹⁶ The effective concentrations can differ vastly, depending on physical parameters such as pH and water hardness. Higher amounts of Ca and Mg in hard water lakes compete with toxic metal ions for binding sites on an organism’s surface, usually decreasing their toxicity.^229,230 This does not apply, however, to copper toxicity, because transporters for Cu²⁺ have such a low affinity for Ca²⁺ and Mg²⁺ that the latter metal cannot outcompete Cu²⁺. Therefore, water hardness does not protect against copper toxicity.²³¹

Conclusions

In most cases of toxicity assays, not only high metal concentrations but also short exposure times were used. It is obvious that at high metal concentrations, the toxicity becomes less specific (metal binding to low-affinity sites once the high-affinity sites are occupied). However, a very recent study on Ni²⁺ toxicity showed that on the basis of biotic ligand models, chronic toxicity cannot be predicted by models for acute toxicity,²³² confirming earlier studies on zooplankton with other metals. To unravel the mechanisms of metal toxicity, it is important to study the effects under environmentally relevant conditions to ensure a specific effect and not an overall inhibition of the metabolism.

We summarized the mechanisms described in this review in the scheme shown in Fig. 1.

Acknowledgements

We would like to thank the Academy of Sciences of Czech Republic for financial support.

Footnotes

†

We generally refer to the element name only, when the redox state is unknown or at the beginning of sentences. In biological systems, Cd, Hg, Ni and Zn have the redox state 2+, while Fe and Cu can be Fe^2+/3+, Cu^+/2+.