Molecular mechanisms of metal hyperaccumulation in plants

Authors

  • Nathalie Verbruggen,

    1. Laboratoire de Physiologie et de Génétique Moléculaire des Plantes, Université Libre de Bruxelles, Campus Plaine – CP242 – Bd du Triomphe, B-1050 Brussels, Belgium;
    Search for more papers by this author
  • Christian Hermans,

    1. Laboratoire de Physiologie et de Génétique Moléculaire des Plantes, Université Libre de Bruxelles, Campus Plaine – CP242 – Bd du Triomphe, B-1050 Brussels, Belgium;
    Search for more papers by this author
  • Henk Schat

    1. Institute of Molecular and Cellular Biology, Faculty of Earth and Life Sciences, Vrije Universiteit, De Boelelaan 1085, 1081 HV, Amsterdam, the Netherlands
    Search for more papers by this author

Author for correspondence:
Nathalie Verbruggen
Tel:+32 2 650 21 28
Fax: +32 2 650 54 21
Email: nverbru@ulb.ac.be

Abstract

Contents

  • Summary 759

  • I. Hyperaccumulation: the phenomenon 759
  • II. Macroevolution of hyperaccumulation 760
  • III. Microevolution of hyperaccumulation: variation within hyperaccumulator species 760
  • IV. Genetic analysis of trace metal accumulation and tolerance 761
  • V. Mechanisms of trace metal accumulation 762
  • VI. General discussion and research perspectives 769
  • Acknowledgements 772

  • References 772

Summary

Metal hyperaccumulator plants accumulate and detoxify extraordinarily high concentrations of metal ions in their shoots. Metal hyperaccumulation is a fascinating phenomenon, which has interested scientists for over a century. Hyperaccumulators constitute an exceptional biological material for understanding mechanisms regulating plant metal homeostasis as well as plant adaptation to extreme metallic environments. Our understanding of metal hyperaccumulation physiology has recently increased as a result of the development of molecular tools. This review presents key aspects of our current understanding of plant metal – in particular cadmium (Cd), nickel (Ni) and zinc (Zn) – hyperaccumulation.

I. Hyperaccumulation: the phenomenon

High tolerance to trace elements (including metals or metalloids) has evolved in a number of plant species. Tolerant plants are often excluders, limiting the entry and root-to-shoot translocation of trace metals. However, a class of rare plants called hyperaccumulators combines extremely high tolerance to, and foliar accumulation of, trace elements. Hyperaccumulators have recently gained considerable interest because of their potential use in phytoremediation (Chaney et al., 1997, 2005; reviewed by Pilon-Smits, 2005), phytomining (Li et al., 2003) and food crop biofortification (Broadley et al., 2007; Palmgren et al., 2008).

Hyperaccumulators can accumulate exceptional concentrations of trace elements in their aerial parts without visible toxicity symptoms. Foliar concentration thresholds for hyperaccumulation are summarized in Table 1. The most extreme example reported to date is probably the latex of Sebertia acuminata, which can contain up to 26% (w/w) Ni (Jaffréet al., 1976).

Table 1.  Hyperaccumulators of trace metals (adapted from Reeves & Baker, 2000; A. Baker pers. comm.; except for cadmium: Bert et al., 2002; Yang et al., 2004; Vogel-Mikušet al., 2005)
Trace elementConcentration criterion (% in leaf dry matter)Number of taxaNumber of families represented
  1. Numbers in brackets refer to unconfirmed data (A. Baker, pers. comm.).

Antimony≥ 0.1  2 2
Cadmium≥ 0.01  4 2
Cobalt≥ 0.1(26)11
Copper≥ 0.1(35)15
Lead≥ 0.1 14 7
Manganese≥ 1.0 10 6
Nickel≥ 0.139042
Thallium≥ 0.1  1 1
Zinc≥ 1.0 14 6

Over 450 plant species (0.2% of angiosperms) have been identified as hyperaccumulators of trace metals (Zn, Ni, Mn, Cu, Co and Cd), metalloids (As) and nonmetals (Se), the majority of them being Ni hyperaccumulators (75%) (Brooks et al., 1974, 1977; Baker & Brooks, 1989; Reeves & Baker, 2000; Ellis & Salt, 2003; Reeves, 2003, 2006; Sors et al., 2005; Milner & Kochian, 2008) (Table 1). As previously noted by Macnair (2003), the phenomenon of hyperaccumulation has been overreported. For example, Cu and Co hyperaccumulation in a number of species from the copper belt in Congo appeared to be the result of airborne contamination of the leaf surface, rather than root uptake and translocation (Faucon et al., 2007).

II. Macroevolution of hyperaccumulation

Hyperaccumulators occur in over 34 different families. The Brassicaceae family is relatively rich in them, in particular the genera Alyssum and Thlaspi. Since Ni hyperaccumulation occurs in a broad range of unrelated families, it is certainly of polyphyletic origin (Macnair, 2003). A relatively large fraction of the Ni hyperaccumulators described so far belongs to the Odontarrhena section in the genus Alyssum (Brassicaceae). Within that section, Mengoni et al. (2003) established that the nonhyperaccumulators are distributed over different clades and that the trait might have been lost or, less likely, acquired more than once. With few exceptions, among them Sedum alfredii (Crassulaceae), Zn and Cd hyperaccumulation is confined to the Brassicaceae, in particular to the genus Thlaspi and the species Arabidopsis halleri. Within the Thlaspi genus, hyperaccumulation is confined to the Noccaea section, and all the members of this section are hyperaccumulators, suggesting that hyperaccumulation is monophyletic within this genus (Macnair, 2003). Some species or populations of Thlaspi hyperaccumulate Ni in their natural (serpentine) environment, but a number of these are capable of Zn hyperaccumulation, whereas Alyssum hyperaccumulators do not hyperaccumulate Zn (Assunção et al., 2001, 2008; Taylor & Macnair, 2006).

How did hyperaccumulation evolve? The selective factors causing the evolution of hyperaccumulation are unknown and difficult to identify retrospectively. The different nonmutually exclusive hypotheses in the literature are: increased metal tolerance, protection against herbivores or pathogens, inadvertent uptake, drought tolerance, and allelopathy (Boyd & Martens, 1992). The hypothesis of protection against herbivores and pathogens is certainly the most popular one (Boyd & Martens, 1994; Pollard & Baker, 1997; Jhee et al., 1999; Huitson & Macnair, 2003; Noret et al., 2007, reviewed in Boyd, 2007; Galeas et al., 2008).

III. Microevolution of hyperaccumulation: variation within hyperaccumulator species

There is a huge variability between populations of hyperaccumulator species in their capacity to tolerate and accumulate metals, which provides opportunities to analyze the underlying genetic determinants. Such variation has been extensively reported for Thlaspi caerulescens (Table 2) and Arabidopsis halleri, which are the two most studied hyperaccumulator species to date. In T. caerulescens, all populations hyperaccumulate Zn, but when grown on the same substrate, populations from metallicolous soils generally accumulate less Zn than populations from nonmetallicolous soils (Meerts & Van Isacker, 1997; Escarréet al., 2000; Dubois et al., 2003; Frérot et al., 2003). Also in A. halleri, Zn hyperaccumulation is constitutive, but there is heritable variation in degree between local populations. However, on average, there is no difference between metallicolous and nonmetallicolous populations under controlled conditions (Bert et al., 2002; Macnair, 2002). High Zn tolerance seems to be constitutive in both species, but again there is considerable variation in degree, with the highest tolerances found in metallicolous populations (Bert et al., 2000; Assunção et al., 2003a,c; Pauwels et al., 2006). When grown in Zn-enriched soil, nonmetallicolous plants show reduced performance, compared with metallicolous plants, demonstrating that the basic degrees of Zn tolerance in nonmetallicolous plants are insufficient to sustain maximum performance in metalliferous soil (Meerts & Van Isacker, 1997; Escarréet al., 2000; Jiménez-Ambriz et al., 2007; Dechamps et al., 2008). As evidenced by molecular markers, the enhanced Zn tolerance in metallicolous populations are of polyphyletic origin, resulting from independent local microevolutionary adaptation (Pauwels et al., 2006; Jiménez-Ambriz et al., 2007).

Table 2.  Foliar metal concentrations in Thlaspi caerulescens populations
 La Calamine (µmol g−1 DW)LellingenGangesMonte Prinzera
  1. Plants were grown for 4 wk in modified Hoagland's solution, of which the last 3 wk with 10 µm Zn, 5 µm Cd or 10 µm Ni (data from Assunção et al., 2003a).

Zn6.528.060.553.9
Cd1.817.329.94.5
Ni1.517.316.948.3

For Cd accumulation, intraspecies variation seems considerably higher than for Zn (Bert et al., 2002; Assunção et al., 2003a,b,c). Cd hyperaccumulation has been reported in only four species, which are also Zn hyperaccumulators, suggesting that Cd and Zn hyperaccumulation relies on common genetic determinants, at least partially (Table 1). In S. alfredii, both Zn and Cd hyperaccumulation are not constitutive at the species level, but confined to metallicolous populations (Yang et al., 2006; Sun et al., 2007). In T. caerulescens the variation in Cd accumulation among local populations is correlated with the variation in Zn accumulation, at least in some studies (Assunção et al., 2003c; Roosens et al., 2003). Xing et al. (2008) observed a strong correlation between the root-to-shoot translocation rates of Cd and Zn among a large number of T. caerulescens and T. praecox populations. These studies confirm the hypothesis of common determinants for Cd and Zn hyperaccumulation. However, Cd and Zn accumulation are uncorrelated among T. caerulescens populations from southern France, where the metallicolous ones accumulate more Cd but less Zn than the nonmetallicolous populations, when grown in the same soil (Escarréet al., 2000). In T. caerulescens, some populations accumulate less Cd than the nonhyperaccumulator congeneric species, T. arvense, irrespective of the Cd supply. However, all the T. caerulescens populations exhibit higher Cd concentrations in leaves than in roots (Assunção et al., 2003c). By contrast, A. halleri, in which Cd hyperaccumulation is not constitutive at the species level, accumulates more Cd in roots than in shoots (Bert et al., 2003). In T. caerulescens, Cd hypertolerance is not constitutive at the species level, but confined to populations from calamine soils. Some exceptional populations from southern France (the region around the village of Ganges) even require Cd for optimal growth (Roosens et al., 2003; Liu et al., 2008), suggesting a possible biological role for Cd in those populations as demonstrated in the marine diatom Thalassiosira weissflogii (Lee et al., 1995; Lane & Morel, 2000; Lane et al., 2005). Thus, in T. caerulescens, enhanced root-to-shoot Cd transport seems to be constitutive, but Cd hypertolerance and high-level Cd hyperaccumulation seem to result from recent microevolutionary change at a local scale.

Like Cd hyperaccumulation, Ni hyperaccumulation does not seem to be constitutive in T. caerulescens but is confined to serpentine populations (Assunção et al., 2003c). Enhanced Ni tolerance may be constitutive at the species level although variation exists among populations. (Assunção et al., 2003c). Under equimolar supply of Zn and Ni, serpentine populations of T. caerulescens, T. pindicum and T. alpinum prefer Zn over Ni (Assunção et al., 2001, 2008; Taylor & Macnair, 2006), suggesting that Ni hyperaccumulation results from local microevolutionary adaptation, and that hyperaccumulation of Zn, rather than of Ni, may be the basic condition in these species. The Turkish serpentine endemics T. oxyceras, T. rosulare and T. violascens, on the other hand, don't seem to prefer Zn over Ni (Peer et al., 2003), suggesting that different strategies for Ni hyperaccumulation may have been evolved within the Noccaea clade.

IV. Genetic analysis of metal accumulation and tolerance

In this section we aim to review our current knowledge of the genetic architecture of metal (Zn, Ni and Cd) accumulation and tolerance.

Zn hyperaccumulation and hypertolerance are to a large extent constitutive at the species level in A. halleri and T. caerulescens, which is a major handicap in the genetic analysis of these traits. Macnair et al. (1999) were the first to circumvent this handicap by analyzing an interspecific cross between A. halleri and A. lpetraea. It may be expected that major hyperaccumulation or tolerance loci that control the segregation in interspecific A. halleri × A. l. petraea crosses may not do so in intraspecific T. caerulescens crosses. This may not be true for nonconstitutive traits, such as Cd or Ni hyperaccumulation and Cd tolerance (see earlier).

From an analysis of a F2 generation of the A. halleri × A. lyrata subsp. petraea cross, Macnair et al. (1999) concluded that Zn tolerance and Zn hyperaccumulation were under independent genetic control. Another progeny from an A. halleri×A. l. petraea F1 plant backcrossed to A. l. petraea (called BC1) was used by Bert et al. (2003) to analyze the co-segregation of Cd tolerance and Cd accumulation. They found independent inheritance of these characters, but a degree of co-segregation of Cd tolerance with Zn tolerance, and Cd accumulation with Zn accumulation, suggesting partial pleiotropic control of these characters. The same BC1 has been used to map quantitative trait loci (QTLs) for Zn and Cd tolerance (Courbot et al., 2007; Willems et al., 2007). Three additive QTLs were found for Zn tolerance and for Cd tolerance, all of them with the trait-enhancing alleles originating from the A. halleri parent (Table 3). One QTL for Cd tolerance co-located with a Zn tolerance locus (Table 3), confirming the hypothesis of partial pleiotropic control of Zn and Cd tolerance (Bert et al., 2003). Filatov et al. (2007), using an F2 generation of the A. halleri×A. l. petraea cross made by Macnair et al. (1999), mapped two QTLs for Zn accumulation to different linkage groups.

Table 3.  Summary of quantitative trait loci (QTL) analysis
CrossTraitQTLLGLOD%PVETrait-enhancing allele
  1. (a) QTLs for zinc (Zn) or cadmium (Cd) accumulation in roots and shoots in crosses between plants from different Thlaspi caerulescens populations. Plants were grown in modified Hoagland's with 2 µm Zn and 5 µm Cd (LC × GA) or 10 µm Zn (LE × LC) (data from Assunção et al., 2006; Deniau et al., 2006). Co-locating QTLs have been marked with the same superscripted letters (LE, Lellingen; LC, La Calamine; GA, Ganges; S, shoot; R, root; PVE, percentage of phenotypic variance explained; LG, linkage group).

  2. (b) QTLs for Zn (Willems et al., 2007) and Cd (Courbot et al., 2007) tolerance in a BC1 backcross progeny of an Arabidopsis halleri×A. lyrata subsp. petraea cross (Ah × Al). ZnT/CdT, Zn/Cd tolerance; further as above.

(a)
LE × LCZnRZnR134.621.7LE
LE × LCZnRZnR253.616.6LC
LC × GAZnSZnS1b16.014.7LC
LC × GAZnSZnS244.0 9.6LC
LC × GAZnSZnS377.418.1GA
LC × GAZnRZnR3b34.014.9LC
LC × GAZnRZnR4c68.554.4GA
LC × GACdSCdS1a35.523.8GA
LC × GACdRCdR1a37.3 9.6GA
LC × GACdRCdR2c68.733.1GA
(b)
Ah × AlZnTZnT1a36.512.2Ah
Ah × AlZnTZnT247.311.2Ah
Ah × AlZnTZnT364.5 5.6Ah
Ah × AlZnSCdT1a39.942.9Ah
Ah × AlZnSCdT244.523.7Ah
Ah × AlZnRCdT364.415.9Ah

Zha et al. (2004), using an F2 generation of an intraspecific T. caerulescens cross between two calamine plants, from Prayon (PR) and southern France (GA), the latter with much higher Cd accumulation capacity, analyzed the segregation and co-segregation of Cd and Zn accumulation and Cd tolerance. Zn, Cd and Mn accumulation in the F2 were phenotypically correlated. They found transgressive segregation of Zn accumulation, but not of Cd accumulation. QTLs for Zn and Cd accumulation have been mapped in two T. caerulescens intraspecific crosses (Table 3). Assunção et al. (2006) found two QTLs for root Zn accumulation in an F3 cross between plants from a Belgian calamine site (LC) and from a nonmetalliferous site in Luxemburg (LE), the latter with higher Zn accumulation capacity. Deniau et al. (2006) mapped QTLs for Cd and Zn accumulation in an F2 cross between two plants from GA and LC, the latter with lower Cd accumulation capacity. There was one common locus for Cd and Zn in root, one for Zn in root and Zn in shoot, and one for Cd in root and Cd in shoot. The trait-enhancing alleles at the three Cd accumulation loci were all derived from the GA parent. In the two interspecific crosses, both of the parents contributed trait-enhancing alleles to the Zn accumulation loci. In agreement with this, Deniau et al. (2006) found transgressive segregation for Zn accumulation, but not for Cd accumulation. The Cd and Zn accumulation rates in the F2 plants were moderately, but significantly phenotypically correlated, in agreement with the hypothesis that Cd and Zn accumulation, in so far as they segregated, are partly governed by common genetic determinants, but additionally by more metal-specific determinants.

Richau & Schat (2008) analyzed the segregation of Zn and Ni accumulation in F3 and F4 families derived from a cross between T. caerulescens from a calamine site (LC) and from a serpentine site at Monte Prinzera, Italy (MP), the latter showing much higher Zn and Ni accumulation capacity. The segregation of both Ni and Zn accumulation was not transgressive in this cross and Zn and Ni accumulation in the F3 was phenotypically correlated. Partitioning of the phenotypic correlation showed that the correlation was entirely genetic, implying that Zn and Ni accumulation, in so far as they segregated in this cross, are pleiotropically controlled by the same genes. This is in agreement with the Zn-suppressible Ni hyperaccumulation in MP (Assunção et al., 2001, 2008).

The genetics of metal hypertolerance have been poorly studied in T. caerulescens. Zn tolerance segregated in LC × LE F2 crosses, but was uncorrelated with Zn accumulation (Assunção et al., 2003b). Similarly, Zha et al. (2004) and Richau & Schat (2008) did not find any phenotypic correlation between Cd accumulation and Cd tolerance, or between Ni accumulation and Ni tolerance, respectively. Thus, metal tolerance and metal accumulation, in so far they segregate in intraspecific T. caerulescens crosses, seem to be under nonpleiotropic genetic control.

The high synteny between the genetic maps of A. halleri and A. thaliana facilitated the identification of candidate genes at QTLs for Zn and Cd tolerance. Those genes will be discussed in the following section.

V. Mechanisms of trace metal hyperaccumulation

Physiological studies have paved the way for a basic understanding of metal hyperaccumulation mechanisms, including enhanced metal uptake, increased xylem loading and increased detoxification in the shoot (Lasat et al., 1996, 1998; Krämer et al., 1996, 1997; Lombi et al., 2001; Zhao et al., 2002, 2006; Assunção et al., 2003a; Xing et al., 2008).

Analyzing trace metal tolerance and accumulation has been greatly enhanced by the use of molecular resources developed for A. thaliana. High-throughput technologies, in particular microarray, have allowed the complexity of the hyperaccumulation phenomenon to be tackled. These studies support the idea that genes that are thought to be involved in hyperaccumulation and hypertolerance are not species-specific or novel, but rather differently expressed and regulated, compared with nonhyperaccumulator species. However as no complete genome sequences of hyperaccumulators are yet available, this assumption cannot be fully verified. Comparative transcriptomics studies on hyperaccumulators and related nonaccumulating nontolerant species have identified a large array of genes that are constitutively (in the absence of excess of metallic ions) highly expressed (Weber et al., 2004, 2006; Becher et al., 2004; Craciun et al., 2006; Filatov et al., 2006; van de Mortel et al., 2006; Talke et al., 2006). These molecular insights are for the moment restricted to species related to A. thaliana. But the ever-growing availability of full-genome sequences will allow the development of microarrays in other plant families.

Hyperaccumulation of Zn is probably the best-understood example of metal hyperaccumulation at the molecular level. The existence of species closely related to A. thaliana displaying Zn hyperaccumulation has allowed rapid progress. The most closely related hyperaccumulator species is A. halleri. These species diverged c. 5 million yr ago and they show c. 94% nucleotide identity within coding regions. Genes differentially expressed in A. halleri compared with A. thaliana (Becher et al., 2004; Weber et al., 2004, 2006; Talke et al., 2006) or in progenies from crosses between A. halleri and A. l. petraea segregating for Zn hyperaccumulation (Filatov et al., 2006) have recently been reviewed by Roosens et al. (2008). Similar studies have taken place in T. caerulescens (Hammond et al., 2006; van de Mortel et al., 2006, 2008). T. caerulescens and A. thaliana diverged c. 20 million yr ago and share c. 88% nucleotide identity in coding regions (van de Mortel et al., 2006). In total, > 2000 genes were found to be at least five times significantly more expressed in T. caerulescens compared with Arabidopsis, of which 1147 had an unknown function. To pinpoint genes directly involved in Zn homeostasis, only those regulated by Zn supply (deficiency or excess) in Arabidopsis and/or in hyperaccumulators were used to compose a shortlist of candidate genes (van de Mortel et al., 2006; Talke et al., 2006). Striking convergence was observed between the genes identified in the two Brassicaceae. Many of them can be categorized in functions associated with metal homeostasis, while others take part in basic metabolism, without direct relation to metal tolerance or accumulation (Table 4).

Table 4.  List of genes more expressed in Arabidopsis halleri (references 1, 2, 3, 5) and/or Thlaspi caerulescens (references 4, 6, 8, 9) compared with nontolerant nonaccumulator relatives (A. thaliana or T. arvense)
Related functionNameAnnotationOrganReferences
  1. Gene names refer to A. thaliana. 1, Becher et al. (2004); 2, Weber et al. (2004); 3, Filatov et al. (2006); 4, Hammond et al. (2006); 5, Talke et al. (2006); 6, van de Mortel et al. (2006); 7, Gendre et al. (2007); 8, van de Mortel et al. (2008). When high gene expression was observed in only one of the two species, the name of the species is added (Ah, A. halleri; Tc, T. caerulescens).*, no transcript in A. halleri shoot except under Zn-deficiency conditions.

Metal uptake into cellsZIP4ZIP family of metal transportersR + S1, 2, 4, 6
ZIP6ZIP family of metal transportersR + S1, 3, 4, 5
ZIP9ZIP family of metal transportersR + S (*)1, 2, 4, 5, 6
ZIP10ZIP family of metal transportersR (lower in S)4, 5, 6
IRT1ZIP family of metal transportersR1, 6
IRT3ZIP family of metal transportersR + S1, 4, 5
ZIP7ZIP family of metal transportersS4
Metal vacuolar sequestrationMTP1Cation diffusion facilitatorR + S1, 2, 5, 6
MTP8Cation diffusion facilitatorR + S4, 5, 6
MPT11Cation diffusion facilitatorS5
CAX2Ca2+: cation antiporter R + S1, 2, 4
AtHMA3P-type metal ATPaseS1, 3, 4, 6
Metal remobilization from the vacuoleNRAMP3Natural resistance associated macrophageR + S (Ah)2, 3, 5
NRAMP1Natural resistance associated macrophageS4, 6
NRAMP5Natural resistance associated macrophageS4, 6
Xylem loading/unloading of metal/ligands/metal–ligand complexesHMA4P-type metal ATPaseS1, 2, 3, 4, 5, 6
FRD3Multidrug and toxin efflux family transporterR + S (Ah)4, 5, 6, 8
YSL3Yellow-stripe-like transporterR + S7
YSL6Yellow-stripe-like transporterS5
YSL7Yellow-stripe-like transporterS7
Synthesis of metal ligandsNAS1Nicotinamine synthase 4, 6
NAS2Nicotinamine synthaseR2, 5, 6
NAS3Nicotinamine synthaseS (Ah) + R(Tc)1, 4, 5, 6
NAS4Nicotinamine synthaseR + S (Tc)2, 4, 6, 8
SAMS1S-adenosyl-methionine synthetaseS5
SAMS2S-adenosyl-methionine synthetaseS5
SAMS3S-adenosyl-methionine synthetaseR + S5
AOSA2Cysteine synthaseR + S2, 5
Other roles in iron homeostasisFER1Ferritin Fe(III) bindingS5, 6
FER2Ferritin Fe(III) bindingS5
IREG2Iron regulated transporter 2R5, 6
At4g35830Cytoplasmic aconitaseS3
Stress protection/responsePDI1Protein disulfide isomerase 1S5
PDI2Protein disulfide isomerase 2S5
At1g45145H-type thioredoxinS3
Homeostasis of macronutrientPHT1-4Phosphate:H1 symporter familyR + S1, 4, 5

Comparative root transcriptome analyses between the same hyperaccumulators and A. thaliana or A. l. petraea were also performed in response to Cd2+ exposure with the aim to identify genes that are important for adaptation to this toxic heavy metal (Craciun et al., 2006; Weber et al., 2006; van de Mortel et al., 2008). Cd is considered as a major environmental pollutant, which enters the food chain mainly by plant uptake. Therefore, it is important to identify the genes responsible for the accumulation of Cd in the different organs of plants to be able to control this trait. There was a constitutive overexpression of genes involved in stress response in A. halleri and T. caerulescens, as well as of genes involved in metal homeostasis. The study in T. caerulescens emphasized the potential role of genes involved in lignin, glutathione and sulfate metabolism.

In short, transcriptomics studies have confirmed the importance of metal transport and detoxification processes in hyperaccumulation but have also shed light on other putative modifications of cellular processes (which will be addressed in the last paragraph of this section).

We will now review our current understanding of key steps involved in hyperaccumulation, illustrated in Fig. 1.

Figure 1.

Overview of our current understanding of adaptations to hyperaccumulate metals, in particular zinc (Zn). Those adaptations are highlighted in red. (I) Enhanced Zn uptake into root cells is thought to be driven by ZIP4 and to some extent by IRT1. ZIP3, ZIP9, and ZIP10 may also play a role, yet to be established. In T. caerulescens a peri-endodermal thickening (apoplastic barrier along the root axis) was observed. (II) Active xylem loading depends on reduced vacuolar root sequestration (main Zn storage is thought to be in the cortex) and enhanced activity of HMA4 in stelar parenchyma, leading to efficient efflux of Zn from symplasm. Efflux of citrate driven by FRD3 seems to play a role in Zn translocation. Loading of Zn in the xylem can also occur as Zn-nicotianamine complex by YSL proteins. (III) In the leaves, Zn is unloaded in vessel-associated cells by YSL proteins, and possibly by FRD3 in Arabidopsis halleri. Enhanced uptake of Zn in shoot symplastic pathway may be favored by ZIP6 in A. halleri. (IV) Detoxification is assumed to be operated by chelation of metals. < > refers to chelation. Possible ligands of Zn in the cytoplasm are histidine and nicotianamine. (V) Vacuolar sequestration in the leaves is thought to be the main pathway of detoxification of metals. Zn is mainly stored in vacuoles of mesophyll (A. halleri) and/or epidermal cells (Thlaspi caerulescens), through the activity of MTP1, and possibly HMA3 and MHX. In the vacuoles, a large pool of malate favors the formation of Zn-malate complexes. (VI) Other adaptive processes include homeostasis of other nutrients, in particular Fe (role of FER1, FER2, IREG2, NRAMP3 in A. halleri and T. caerulescens; NRAMP1,5 in T. caerulescens); P (PHT1-4), enhanced stress responses/protection (higher GSH level, higher level of defensins, etc.).

1. Uptake of metal from the soil to the root

Metal avaflability and mobility in the rhizosphere can be influenced by root exudates, such as siderophores, organic acids and protons, as well as by rhizosphere microorganisms (Zhao et al., 2001; Whiting et al., 2001; Wenzel et al., 2003). However, there is no definite answer to the question of whether, and how, hyperaccumulators and nonhyperaccumulators, or their root-associated microbial communities, have different effects on the metal availability in their rhizospheres. In T. caerulescens, high metal uptake has been associated with enhanced root proliferation in artificially metal-enriched soil patches under experimental conditions (Whiting et al., 2000). However, not all T. caerulescens populations exhibit this metal root foraging (Haines, 2002), and its relevance under natural conditions is unknown.

Enhanced Zn root uptake seems to be driven by overexpression of members of the ZIP family (zinc-regulated transporter, iron-regulated transporter protein; reviewed by Krämer et al., 2007) (Fig. 1). Early work by Pence et al. (2000) has highlighted the constitutive overexpression in T. caerulescens roots of ZNT1 resulting in root Zn uptake with higher Vmax (but similar Km) than in the related nonaccumulator species, T. arvense, as observed in a previous radiotracer flux study (Lasat et al., 1998). ZNT1 mediates high-affinity Zn transport as well as low-affinity Cd uptake (Pence et al., 2000) and is the homolog of AtZIP4. High expression of ZNT1 was later confirmed in microarray analyses (Hammond et al., 2006; van de Mortel et al., 2006). Many Zn-transporting ZIP members, including ZNT1, are Zn-regulated in nonaccumulators, that is, they are only detectably expressed under conditions of Zn deficiency, whereas they are expressed more or less independently of the Zn supply in hyperaccumulators (Pence et al., 2000; Assunção et al., 2001).

Physiological studies on T. caerulescens have provided strong evidence that multiple uptake systems are involved in the root uptake of Cd and Zn (Lombi et al., 2002, Zhao et al., 2002; Cosio et al., 2004, Roosens et al., 2004). Those include a system with a strong preference for Zn over Cd, and another one with a preference for Cd over Zn. Based on physiological evidence, the high rate of Cd hyperaccumulation in the populations from southern France (Lombi et al., 2002; Cosio et al., 2004; Roosens et al., 2004) is attributable to the expression of a Cd-preferent accumulation system, which also accumulates Zn when there is a lot more Zn than Cd in the nutrient solution (Zhao et al., 2002). In the populations from northern Europe, Zn accumulation is predominantly due to a Zn-preferent accumulation system (Zhao et al., 2002). The transgression of Zn accumulation found in the interspecies crosses between calamine populations of Prayon or La Calamine (B) and Ganges (southern France) (see Section IV) would then be explained by independent recombination of the genes controlling both systems.

Studies of metal influx showed that Vmax for Cd uptake was c. five times higher in the Ganges population than in Prayon; Km was similar, and Cd uptake was enhanced by Fe deficiency in Ganges but not in Prayon (Lombi et al., 2001). Cd was able to compete with Fe uptake in populations from the Ganges region but not in Prayon (Roosens et al., 2003). In the latter population, Ca uptake was inhibited by Cd addition, suggesting that, in addition to two Zn uptake systems, at least two other uptake systems are involved in Cd uptake (Roosens et al., 2003). IRT1 has been suggested to be responsible for Cd hyperaccumulation in the Ganges population (Lombi et al., 2001). However, in contrast to AtIRT1, TcIRT1 seems to be incapable of Cd transport (Plaza et al., 2007).

Microarray analyses have highlighted the overexpression of more ZIP members in A. halleri and T. caerulescens (the homologs of AtZIP3, AtZIP6, AtZIP9, AtZIP10 and AtIRT3), but their roles in plants and in Zn hyperaccumulation remain to be established (Becher et al., 2004; Weber et al., 2004, 2006; Filatov et al., 2006; Hammond et al., 2006; van de Mortel et al., 2006, 2008; Talke et al., 2006; reviewed in Krämer et al., 2007). Candidate transporters responsible for Ni hyperaccumulation in serpentine T. caerulescens have not been identified. Several Ni-hyperaccumulating populations prefer Zn over Ni in experimental conditions (Assunção et al., 2001), suggesting that Ni is taken up by a Zn transporter in these populations. However, in other populations, there may be Ni-preferent transporters (Peer et al., 2003).

2. Translocation from the root to the shoot

Efficient translocation of metal ions to the shoot requires radial symplastic passage and active loading into the xylem (Clemens, 2006; Xing et al., 2008). The need of ligands for all trace metals in the xylem is controversial. Salt et al. (1999) showed that most Zn in the xylem sap of T. caerulescens was present as the free hydrated Zn2+ ion. Similarly, Ueno et al. (2008) showed that Cd occurred mainly in the free ionic form in the xylem sap in A. halleri.

Availability of trace metals for translocation to the shoot implies limited sequestration in vacuoles of root cells (Fig. 1). Lasat et al. (1998) observed a much lower accumulation of Zn in root vacuoles, as well as a faster release of Zn from the root vacuoles in T. caerulescens, compared with T. arvense. Yang et al. (2006) reported that in the nonaccumulating ecotype of S. alfredii, 2.7-times more Zn was retained in the root vacuoles than in the hyperaccumulating ecotype. Recently, Xing et al. (2008) reported variation in root vacuolar Cd fractions, associated with different translocation efficiencies, between two T. caerulescens populations.

Physiological studies of hyperaccumulators also demonstrated higher metal concentrations in the xylem sap due to enhanced xylem loading (Lasat et al., 1998). Several types of transporters are involved in this process.

P-type ATPase-HMA  Molecular studies and mutant analysis have identified particular P-type ATPases as being responsible for the Cd and Zn loading of the xylem from the surrounding vascular tissues. The P1B-type ATPases, also known as the heavy metal transporting ATPases (HMAs), play an important role in transporting transition metal ions against their electrochemical gradient using the energy provided by ATP hydrolysis. In bacteria, P1B ATPases are the main players in metal tolerance (Monchy et al., 2007). HMAs cluster into two classes: those transporting monovalent cations (Cu/Ag group) and those transporting divalent cations (Zn/Co/Cd/Pb).

HMA4 was the first gene encoding a plant P1B-type ATPase of the divalent transport group to be cloned and characterized in A. thaliana (Mills et al., 2003). HMA4 is localized at the plasma membrane. A role for HMA4 in Zn homeostasis, Cd detoxification, and in the translocation of these metals from the root to the shoot has been demonstrated in A. thaliana, where HMA4 expression seems to be confined to the stele (Mills et al., 2003, 2005; Hussain et al., 2004; Verret et al., 2004, 2005). In both A. halleri and T. caerulescens, HMA4 is more expressed in both roots and shoots compared with Cd/Zn-sensitive close relatives (Bernard et al., 2004; Papoyan & Kochian, 2004; Hammond et al., 2006; van de Mortel et al., 2006; Talke et al., 2006; Courbot et al., 2007), strongly supporting the idea that HMA4 plays an important role in tolerance and/or accumulation of both metals. HMA4 would be involved in cytosolic metal efflux and particularly in root detoxification by translocating Cd and Zn to the shoot. In A. halleri, this hypothesis was reinforced by QTL analysis showing co-localization of a major QTL for Zn and Cd tolerance with the HMA4 gene (Courbot et al., 2007; Willems et al., 2007). HMA4 is currently the only gene for which there is genetic evidence for a role in both Zn and Cd tolerance. High HMA4 expression in the BC1 between A. halleri and A. l. petraea co-segregated with the HMA4 allele originating from the A. halleri parent and with Cd tolerance (Courbot et al., 2007). Recently, Hanikenne et al. (2008) showed that the enhanced HMA4 expression in A. halleri results from triplication of a genomic region containing HMA4, as well as from altered cis-regulation. These authors demonstrated the importance of HMA4 in the hyperaccumulation of Zn and hypertolerance to Cd using RNAi-mediated silencing. A. halleri plants (from a Cd-hypertolerant accession) with a lowered expression of HMA4 translocated less Zn from the root to the shoot (Zn accumulation was shown in the pericycle cells) and were more sensitive to Cd and Zn treatments. Cd translocation was not measured. Interestingly, when expressed in A. thaliana under the activity of its own promoter, AhHMA4 increased Zn and Cd sensitivity because of the absence of a detoxification mechanism with sufficient capacity to cope with the strongly enhanced metal accumulation in the leaves (Hanikenne et al., 2008).

Transcriptomic studies in hyperaccumulators have also revealed a higher expression of genes encoding metal ligands or metal-ligand complex transporters (ligands are reviewed in the following subsection). Some of these transporters seem to play a role in trace metal translocation.

MATE  MATE is a large family of multi-drug and toxic compound extrusion (or efflux) membrane proteins. Some members of the family were shown to function as drug/cation antiporters that remove toxic compounds and secondary metabolites from the cytosol by exporting them out of the cell or sequestering them to the vacuole (Delhaize et al., 2007). FRD3 is a member of the MATE subfamily, which is thought to efflux citrate into the root vascular tissue. Citrate is necessary for the transport of Fe and possibly also Zn (Durrett et al., 2007). FRD3 is constitutively overexpressed in A. halleri and T. caerulescens compared with A. thaliana and may play a role in Zn translocation (van de Mortel et al., 2006; Talke et al., 2006).

OPT  OPT is a superfamily of oligopeptide transporters including the yellow-stripe 1-like (YSL) subfamily. Some YSL transporters are involved in the loading and unloading of nicotianamine-metal chelates from the vascular tissues. There is evidence for a role of YSL transporters in the Zn and Ni hyperaccumulation of T. caerulescens, especially for TcYSL3 and TcYSL7, which are expressed in xylem parenchyma and phloem (Gendre et al., 2007, reviewed in Haydon & Cobbett, 2007). Furthermore, TcYSL3 was shown to transport Ni-NA chelates (Gendre et al., 2007).

3. Sequestration in the shoot vacuoles

The ability to hyperaccumulate Zn, Ni and Cd seems to be governed, at least in part, by an enhanced capacity of metal storage in leaf vacuoles. Several families of transporters are involved in this process.

CDF  The family of cation diffusion facilitators (CDF) in plants, also called metal transporter proteins (MTPs), contains members involved in the transport of Zn2+, Fe2+, Cd2+, Co2+ and Mn2+not only from cytoplasm to organelles or apoplasm, but also from the cytoplasm to the endoplasmic reticulum (Peiter et al., 2007). ZAT (ZINC TRANSPORTER OF ARABIDOPSIS THALIANA), recently renamed AtMTP1, encodes a Zn transporter involved in vacuolar sequestration in A. thaliana. Overproduction of the Zn transporter ZAT in A. thaliana resulted in higher Zn tolerance and a twofold higher Zn accumulation in roots (Van der Zaal et al., 1999). MTP1 homologs seem to be involved in the Zn hypertolerance trait. In A. halleri, MTP1 was highly expressed, in both roots and shoots, due to expanded copy number (Dräger et al., 2004). Two MTP1 copies co-segregated with Zn tolerance in the BC1 generation of a cross between A. halleri and A. lpetraea (Dräger et al., 2004) and co-localized with two QTLs for Zn tolerance (Willems et al., 2007). In T. caerulescens the AtMTP1 homolog, ZTP1, was highly expressed in leaves and may play a role in vacuolar sequestration too (Assunção et al., 2001).

In the Ni/Zn hyperaccumulator T. goesingense, other CDF members, TgMTP1t1 and TgMTP1t2 (derived from one single copy genomic sequence), were proposed to be involved in Ni vacuolar detoxification (Persans et al., 2001). However a more recent study revealed that TgMTP1 seems to be localized at the plasma membrane, where it could mediate both Ni and Zn efflux from the cytoplasm (Kim et al., 2004).

Other CDF members may play a role in the hypertolerance of other trace metals. ShMTP is involved in the vacuolar storage of Mn in the Mn-hypertolerant tropical legume Stylosanthes hamata. ShMTP conferred higher tolerance and accumulation of Mn when overexpressed in A. thaliana (Delhaize et al., 2003). The role of CDF in Mn hyperaccumulators has not yet been investigated.

HMA  A role of the AtHMA3 homolog in Zn hyperaccumulation was suggested by comparative transcriptome analysis between A. halleri (shoot) or T. caerulescens (root) and A. thaliana or T. arvense (Becher et al., 2004; Hammond et al., 2006; van de Mortel et al., 2006). Yeast expression studies supported a role for AhHMA3 in Zn vacuolar transport (Becher et al., 2004) but not for AtHMA3 (Gravot et al., 2004). The difference in phenotypes can be due to the different strengths of the mutants defective in Zn transport that were used in the two studies. By contrast, the same authors showed evidence for Cd transport by AtHMA3 but not by AhHMA3.

CaCA  In the Ca2+/cation antiporter (CaCA) superfamily, MHX is a vacuolar Mg2+ and Zn2+/H+ exchanger (Shaul et al., 1999): MHX protein was present in the leaves of A. halleri at much higher concentrations than in A. thaliana and was therefore proposed to play a role in Zn vacuolar storage. Since transcript levels were similar in both species, a difference in post-transcriptional regulation was postulated (Elbaz et al., 2006).

Members of other CaCA subfamilies may also play a role in metal detoxification. CAX is the acronym for cation exchanger. It is a large family of membrane proteins, which was recently subdivided into ‘true’ CAX (CAX1–CAX6) and CCX (calcium cation exchanger) (CCX 1–5, previously named CAX 7–11) (Shigaki et al., 2006). Apart from a recent report on CCX3 (Morris et al., 2008), only CAX members of A. thaliana belonging to the true CAX clade have been characterized until now, and all seem to be involved in metal vacuolar sequestration, in particular of Cd. In vitro studies have shown a Cd2+/H+ antiport activity in tonoplast-enriched vesicles from oat roots with a Km of 5.5 µm (Salt & Wagner, 1993). Clemens (2006) questioned the availability of Cd2+ ions in the micromolar range in planta. Nevertheless there is experimental evidence that AtCAX2 and AtCAX4 can transport Cd2+ into the vacuoles. Under the activity of the constitutive 35S CaMV promoter, overexpression of AtCAX2 and AtCAX4 resulted in higher accumulation of Cd into the root vacuoles (Korenkov et al., 2007). Some of the CAX Arabidopsis homologs are highly expressed or differentially expressed in response to Zn in Zn/Cd hyperaccumulators: CAX2 (Becher et al., 2004, Weber et al., 2006) and CAX8/CCX2 in A. halleri (Craciun et al., 2006); CAX3 and CAX7/CCX1 in T. caerulescens (van de Mortel et al., 2006, 2008). However, their metal specificities have not been investigated, and yeast expression experiments suggest that CAX2 does not transport Zn (Becher et al., 2004).

ABC  The superfamily of ABC (ATP-binding cassette) transporters is involved in many physiological processes. Some of the ABC transporters are involved in vacuolar sequestration of various metals or xenobiotics. In two subfamilies, MRP and PRD, members are involved in the transport of chelated heavy metals or the organic acids necessary for the transport of heavy metals. There is strong evidence for a role in trace metal homeostasis (Song et al., 2003; Hanikenne et al., 2005; Kim et al., 2006) and they may be expected to contribute to trace metal hyperaccumulation, in particular to vacuolar sequestration. Two ABC genes were identified in T. caerulescens: the AtMRP10 homolog was shown to be differentially expressed in the shoots of two T. caerulescens populations displaying contrasting Zn tolerance and accumulation (Hassinen et al., 2007) and ATH13 was more expressed in the shoot compared with A. thaliana (van de Mortel et al., 2008). However, direct evidence for a role of these genes in vacuolar sequestration is lacking.

4. Detoxification by chelation of trace metals

It is assumed that most of the hyperaccumulated metals are bound to ligands, such as organic acids, amino acids, peptides and proteins. While some aspects of metal detoxification by ligands have been discovered, there is no complete picture of the different chelators involved in the different stages of the plant-internal transport and storage of metals in hyperaccumulators. Furthermore, the nature of the chelators differs, depending on the location within the plant and the age of the plant (Salt et al., 1999). In T. caerulescens plants, a significant fraction of the Zn within the roots seemed to be associated with histidine, while in the shoot, most Zn was associated with organic acids (Salt et al., 1999). Küpper et al. (2004) found that a large fraction of the foliar Cd was bound to sulfur ligands in T. caerulescens, whereas Ueno et al. (2005) deduced that Cd in the shoot was rather bound to organic acids, mainly malate.

For detailed analysis on ligands, we refer to the reviews of Callahan et al. (2006) and Haydon & Cobbett (2007). We summarize here some of the key ligands that seem to play a role in hyperaccumulation.

Histidine  Histidine (His) is considered to be the most important free amino acid involved in hyperaccumulation (Callahan et al., 2006). It forms stable complexes with Ni, Zn and Cd and it is present at high concentrations in hyperaccumulator roots (Persans et al., 1999). In the Ni hyperaccumulator Alyssum lesbiacum, Ni exposure induced a dose-dependent increase in His in the xylem sap, which was not found in the nonhyperaccumulator congeneric species, A. montanum. Moreover, exogenously supplied His greatly enhanced Ni tolerance and both Ni and His loading into the xylem in A. montanum, but not in A. lesbiacum, probably due to the much higher constitutive root His pool in the latter species (Krämer et al., 1996; Kerkeb & Krämer, 2003). In line with this, Ingle et al. (2005) observed an enhanced expression of the first enzyme of the His biosynthetic pathway, ATP-phosphoribosyltransferase, in A. lesbiacum, as compared with A. montanum. However, His overproducing transgenic A. thaliana lines displayed elevated Ni tolerance, but did not exhibit increased Ni concentrations in xylem sap or in leaves (Wycisk et al., 2004; Ingle et al., 2005). This suggests that His-dependent Ni xylem loading may not be universal in Brassicaceae, and that additional factors are required in at least A. thaliana. The mechanism of His-coupled Ni xylem loading has not been identified yet. Recent results obtained with T. caerulescens suggest that Ni-His complex formation strongly inhibits the retention of Ni in root cell vacuoles (K. H. Richau & H. Schat, pers. comm.).

Nicotianamine  Synthesis of nicotianamine (NA) from 3 S-adenosyl-methionine (SAM) by NA synthase (NAS) is present in all plants. NA forms strong complexes with most transition metal ions. The role of NA seems to be in the movement of micronutrients throughout the plant (Stephan & Scholz, 1993). The study of the semi-lethal tomato chloronerva mutant, deficient in NAS, showed impairment in Fe and Cu distribution (Herbik et al., 1996).

Nicotianamine seems to be involved in metal hyperaccumulation, both in A. halleri and in T. caerulescens, in which several NAS genes showed higher transcript levels. Higher expression of SAM synthetase genes may also be involved in enhanced nicotianamine synthesis (Talke et al., 2006). Higher NA content was observed in A. halleri together with higher AhNAS2 transcript and protein levels in roots (Weber et al., 2004) and higher transcript levels of AhNAS3 in the shoot compared with A. thaliana (Becher et al., 2004). NA was proposed to act as a Zn cytosolic buffer, keeping Zn ions in a detoxified form, available for translocation to the shoot. Microarray analysis in T. caerulescens indicated that NAS1, 3 and 4 are more expressed than in Arabidopsis (Hammond et al., 2006; van de Mortel et al., 2006). In this species a role for NA in Ni hyperaccumulation was proposed and Ni–NA complexes were identified in Ni-exposed roots (Vacchina et al., 2003; Mari et al., 2006). TcNAS1 was constitutively highly expressed in the shoot of T. caerulescens plants (Mari et al., 2006). In response to Ni, NAS was also induced in roots, where it chelated absorbed Ni and facilitated its transport to the shoot. Increase in Ni tolerance could be gained upon NAS overexpression in nontolerant species (Douchkov et al., 2005; Kim et al., 2005; Pianelli et al., 2005). More recent results in Thlaspi hyperaccumulators support the assertion that NA is involved in hyperaccumulation of Ni but not of Zn (Callahan et al., 2007).

Organic acids (citrate, malate)  Because of the low association constants of organic acids with metals, Callahan et al. (2006) argued against a role for organic acids in the hyperaccumulation mechanism (such as long-distance transport), in spite of their constitutively elevated concentrations in hyperaccumulators (Lee et al., 1978; Ueno et al., 2005; Montargès-Pelletier et al., 2008). So their role may be limited to vacuolar sequestration. The formation of metal–organic acid complexes is favored in the acidic environment of the vacuole (Haydon & Cobbett, 2007). A large proportion of Zn in the shoot was associated with malate in A. halleri (Sarret et al., 2002). Ni seems to be mainly associated with citrate in the shoots of Alyssum (Lee et al., 1978) and T. goesingense (Krämer et al., 2000).

Glutathione  Glutathione (Glu-Cys-Gly; GSH) is a major cellular antioxidant. It can form complexes with several metals and is the precursor of phytochelatins. Increased production of glutathione in T. goesingense and other Thlaspi Ni hyperaccumulators is thought to provide protection against oxidative damage due to high Ni concentrations (Freeman et al., 2004). Enhanced glutathione synthesis is driven by constitutive activation of the sulfur assimilation pathway, in particular through enhanced activity of mitochondrial serine acetyltransferase (SATm) (Freeman et al., 2004). The metal tolerance profile of T. goesingense was mimicked in A. thaliana expressing the T.g SATm gene (Freeman & Salt, 2007). Cd exposure also enhanced sulfate and GSH metabolism in T. caerulescens (van de Mortel et al., 2008), and appeared to increase the foliar and root GSH concentrations in a hyperaccumulating S. alfredii population, but not in a nonaccumulating one, where GSH decreased owing to PC synthesis (Sun et al., 2007). These results suggest that there is a role for GSH in hyperaccumulation, which is probably associated with its antioxidant activity.

Phytochelatins  Phytochelatins (PCs) (general formula (GluCys)nGly, where n = 2–11) are synthesized enzymatically from glutathione in the presence of certain metals and metalloids and are ubiquitous in plants (reviewed in Clemens, 2006). Although PCs do have a role in basal metal detoxification, they do not seem to be involved in Cu, Cd, Zn, Co and Ni hypertolerance (Ebbs et al., 2002; Schat et al., 2002; Hernandez-Allica et al., 2006). In hyperaccumulators, just as in nonhyperaccumulators, PCs are mainly induced in the roots, in particular by Cd, but not (or barely) by Zn or Ni, and considerable rates of Cd-induced PC accumulation have only been found in Cd-sensitive, nonmetallicolous or serpentine populations of T. caerulescens and in a nonaccumulating S. alfredii (Schat et al., 2002; Sun et al., 2007). Arsenic, which is normally a very effective inducer of PC synthesis in other species, induces only inconsiderable PC concentrations in the roots of the As hyperaccumulator, Pteris vittata (Zhao et al., 2003). These results suggest that PCs are generally not essential for the hyperaccumulation phenotype.

Metallothioneins  Variation in expression levels of MT family members between plant populations has been associated with variation in Cu tolerance (Murphy & Taiz, 1995; Van Hoof et al., 2001; Jack et al., 2007). MTs of the types 1, 2 and 3 are predominantly regulated by Cu, and seem to function in Cu accumulation and phloem Cu transport (Guo et al., 2008).

Overexpression of several members of the MT family (type 1, 2 and 3) compared with Arabidopsis, and variations in expression levels between populations have been reported for T. caerulescens (Roosens et al., 2004, 2005; Rigola et al., 2006; Hassinen et al., 2007). Type 3 MT was particularly strongly expressed in Cd-hyperaccumulating populations from southern France and was further induced by Cu addition. Several lines of evidence suggested that TcMT3 may be involved in Cu homeostasis (Roosens et al., 2004). Recently, V. I. Hassinen et al. (pers. comm.) studied the co-segregation of various MTs with Zn accumulation in different crosses between T. caerulescens populations with different MT expression levels and found no significant co-segregation.

5. Other adaptive processes in hyperaccumulators

In hyperaccumulators there is evidence for modification of the homeostasis of metals other than the hyperaccumulated ones, such as Cu (Roosens et al., 2004; Talke et al., 2006), Mn (Talke et al., 2006; Krämer et al., 2007) and, in particular, Fe homeostasis (Filatov et al., 2006; Hammond et al., 2006; Talke et al., 2006; van de Mortel et al., 2006). Genes previously associated with iron homeostasis, such as a cytosolic aconitase gene, IRT1, FERRITIN genes (FER1, FER2), NRAMP3, IRON REGULATED TRANSPORTER 2 (IREG2) and FERRIC CHELATE REDUCTASE genes (FRO2), are overexpressed in A. halleri and/or T. caerulescens (Becher et al., 2004; Weber et al., 2004; Filatov et al., 2006; van de Mortel et al., 2006, 2008; Talke et al., 2006). A compelling example is NRAMP3, whose expression co-segregated with the Zn-hyperaccumulation phenotype in the F3 progeny from a cross between A. halleri and A. l. petraea (Filatov et al., 2006). In A. thaliana, NRAMP3 is expressed in the vascular bundles of roots, stem and leaves. AtNRAMP3 is localized on the tonoplast and is proposed to remobilize vacuolar pools of Fe, Cd and Mn (Thomine et al., 2003). The roles of AtNRAMP3 homologs in metal hyperaccumulation are not known. There is no evidence for a Zn transport activity for At or TcNRAMP3 (Thomine et al., 2003; Oomen et al., 2009). In support of a putative role in tolerance, the nramp3nramp4 double mutant of A. thaliana displayed strong hypersensitivity to high Zn and Cd exposures. However, metal tolerance conferred by TcNRAMP expression in the nramp3nramp4 mutant did not exceed that of wild-type A. thaliana (Oomen et al., 2009).

Impact of hyperaccumulation on macronutrient homeostasis is suggested by overexpression of genes predicted to encode K+ transporters and high-affinity phosphate transporters (Hammond et al., 2006; van de Mortel et al., 2006).

Evolution of hyperaccumulation seems to have been accompanied by modification of signals and proteins usually involved in pathogen response. There is a constitutive overaccumulation of salicylic acid (SA) in nickel hyperaccumulators in the Thlaspi genus (Freeman et al., 2005). SA is a key signal involved in plant pathogen response and may thus contribute to the elevated expression of pathogen-responsive genes. Overexpression of defensins/PDF genes was observed in A. halleri and T. caerulescens compared with A. thaliana (Becher et al., 2004; van de Mortel et al., 2006; Talke et al., 2006). A. halleri defensin cDNAs (AhPDF) specifically induced higher Zn tolerance in yeast (Mirouze et al., 2006). Defensins accumulated to a higher degree in A. halleri than in A. thaliana, and transgenic A. thaliana plants overexpressing AhPDF also showed slightly increased tolerance to Zn. The function of defensins is unclear. Plant defensins have usually been associated with an antifungal activity. The current hypothesis is that defensins interfere with divalent metal cation (perhaps Zn2+) channels (Mirouze et al., 2006).

VI. General discussion and research perspectives

1. Hyperaccumulation dissected with physiological and molecular tools

Hyperaccumulators constitute an exceptional biological material and gene reservoir that can be exploited to understand adaptation to extreme metallic environments. Physiological and classical genetic studies have been complemented by molecular studies, in particular transcriptome analysis. Presently it is not always easy to reconcile the results of the different research approaches applied.

Physiological studies  Physiological studies in T. caerulescens have revealed considerable diversity among populations with regard to the capacities and the specific metal-affinity patterns of the accumulation systems, although there may also be a basic system which is common to all the populations (Assunção et al., 2001, 2003a,c, 2008; Zhao et al., 2002). This system is possibly driven by ZNT1 and ZNT2 (the AtZIP4 and AtIRT3 homologs, respectively), which are highly expressed in all T. caerulescens populations investigated so far. As such, this variation is also reflected by genetic studies, yielding a broad segregation of accumulation traits, often with significant transgression in case of Zn. Moreover, QTL analyses of Cd and Zn accumulation yielded QTLs for Cd or Zn exclusively, but also a common one, and, in the case of Zn accumulation, with trait-enhancing alleles originating from different parents. Although intraspecific, these variations are up to one order of magnitude (Table 2). Unfortunately, QTL candidate genes have not been identified yet, primarily because the maps were AFLP-based, which makes it impossible to reconcile results with molecular work. Physiological and genetic analyses of A. halleri × A. lpetraea crosses also yielded consistent results, that is, the co-segregation of Cd and Zn tolerance was explained by a common QTL (Courbot et al., 2007; Willems et al., 2007). QTL analysis in this cross yielded candidate genes underlying the QTLs for Zn and Cd tolerance, in particular HMA4 for Cd and Zn tolerance, and MTP1A and MTP1B for Zn tolerance. In agreement with the QTL analysis, Cd and Zn tolerance (and Zn accumulation) were partly lost in A. halleri upon silencing HMA4 (Hanikenne et al., 2008). To date, this is the only case of consistency among physiological, genetic and molecular results.

Microarray analyses  Microarray analyses have undoubtedly advanced our knowledge of hyperaccumulation by providing promising candidate genes. However, differential expression of genes does not imply that they contribute to metal tolerance or accumulation (Roosens et al., 2008). Differential expression can be at the origin or a consequence of the hyperaccumulation trait; or merely a result of divergence between species. Differential expression can also be the result of adaptation to environmental factors other than excess metals or, generally speaking, to linkage disequilibrium through selection at multiple loci. To distinguish genes directly involved in metal tolerance and/or accumulation from all the other genetic changes, regulation by the hyperaccumulated metal was taken as an additional selection criterion (van de Mortel et al., 2006, 2008; Talke et al., 2006). It is, however, possible that important players are not differentially regulated by Zn and have been lost in that selection. Another filter strategy was to combine genomics and classical genetics approaches and to do comparative transcriptomic analysis in contrasting phenotypes derived from crosses between a hyperaccumulator and a nonhyperaccumulator (Craciun et al., 2006; Filatov et al., 2006). However, the HMA4 gene was not identified in these two studies, which were performed on progenies of crosses between A. halleri and A. l. petraea. In the study by Craciun et al. (2006), a comparative cDNA-AFLP analysis was applied to a cross segregating for Cd tolerance. Because this technique is based on a double digestion, it can generate short fragments, which are lost in the analysis. This was, for example, the case for HMA4. The reason why the gene did not pop up in the microarray analysis by Filatov et al. (2006) is elusive. An additional limitation of interspecies transcriptional studies is an underappreciation of genes with lower expression in hyperaccumulators. For example, twofold lower signal intensity was recorded for T. caerulescens compared with the A. thaliana signal intensities in the study of van de Mortel et al. (2006). Down-regulated genes can be important for hyperaccumulation; for example, down-regulation of genes encoding root vacuolar metal transporters is expected to result in decreased vacuolar metal sequestration in the root, allowing for a higher availability of metals for xylem loading. Finally, some genes can be expressed similarly at the transcript level, while being differentially regulated at the post-transcriptional level, resulting in different protein levels. This has been illustrated by the MHX case (Elbaz et al., 2006).

Despite these drawbacks, a picture emerged for molecular mechanisms underlying hyperaccumulation, in particular for Zn (Fig. 1). The study of hyperaccumulators has further unraveled the role of genes involved in metal homeostasis previously identified in A. thaliana. There is a remarkably convergent core set of genes encoding members of the ZIP, CDF, HMA and NRAMP transporter families, as well as FRD3 and NAS genes overexpressed in Zn hyperaccumulators studied so far. The identification of those genes enables transgenic strategies to engineer plants with higher tolerance capacities or modified accumulation of trace metals. Those transgenic plants could be designed for phytoremediation, biofortification or enhanced food safety (crops with lower accumulation of toxic metals in edible organs). The expectations from plant genetic modifications (PGM) are limited by ever-increasing regulation over PGM, as already mentioned by Baker & Whiting (2002), and green dogmatism.

The study of hyperaccumulators has also provided valuable insights into fundamental mechanisms of metal homeostasis that all plants possess in order to regulate the cellular concentrations of essential and nonessential metal ions. Yet many genes up-regulated in hyperaccumulators are still of unknown function. Nevertheless, at least some light has been shed on the molecular mechanisms of metal-ion uptake processes, chelation, translocation, detoxification and sequestration of trace metals. However, comparative studies with the model plant A. thaliana should be carefully considered because they are based on the assumption that conservation implies function. This assumption does not seem to be always true. For example TcZNT1 proteins are the AtZIP4 homologs but they seem to transport different metals: AtIRT1 can transport Cd in addition to Fe, but the two homologs in T. caerulescens cannot (Plaza et al., 2007).

2. Prospects for further research on hyperaccumulators

Little is known about the evolution of the mutations leading to the observed regulatory changes in the transcriptome. Trait evolution seems to be associated primarily with regulatory changes in gene expression. Bioinformatics methods offer a way to identify promising functional noncoding regions and to narrow the focus for experimental tests (Wray & Babbitt, 2008). Such tools are of limited help without the availability of complete hyperaccumulator genome sequences. Our understanding of hyperaccumulator genome evolution could be greatly improved by merging molecular and ecological genomics. One of the exciting research avenues is the search for signatures of recent adaptive evolution across candidate genes for metal tolerance or accumulation. Strong reductions in silent polymorphism across the genomic neighborhood of candidate genes would indicate whether the target gene has recently experienced fixation of an allele through directional positive selection.

The phenomenon of constitutive overexpression of a large array of genes seems to be a common process in the adaptation of plants to extreme environments. Thellungilla halophila (salt cress) is a close relative of A. thaliana, with which it displays 95% identity at the cDNA level (Inan et al., 2004; Taji et al., 2004; Gong et al., 2005). Salt cress plants are salt-tolerant and can grow in 500 mm NaCl medium, without special morphological alterations. It appeared that a large number of genes that are stress-inducible in A. thaliana were constitutively overexpressed in T. halophila. The possible importance of epigenetics in plant adaptation to extreme environments has not yet been investigated.

Four possible mechanisms to account for the large transcriptional modifications seen in Zn hyperaccumulators have been summarized by Talke et al. (2006):

  • • specific cis-elements or trans-factors controlling transcript levels
  • • deregulation of the Zn sensing machinery
  • • higher Zn requirement
  • • indirect consequences of an alteration in the metal homeostasis network of another element.

Gene duplication (e.g. HMA4, MTP1, ZIP3, ZIP9 in A. halleri and ZIP4/ZNT1-2, and IRT1 in T. caerulescens) may contribute to the ability of hyperaccumulators to express genes at a very high level. In the case of HMA4, modification of cis-regulatory elements contributes to the high expression level of each copy (Hanikenne et al., 2008). The possible importance of trans-acting factors has been little studied. A total of 131 genes encoding putative transcription factors were up-regulated (more than fivefold higher expression) in T. caerulescens compared with A. thaliana (van de Mortel et al., 2006) but no clear candidates have popped up (van de Mortel et al., 2008).

A deregulation of the root Zn deficiency response (Becher et al., 2004; Talke et al., 2006; Weber et al., 2004, 2006) may be related to the higher Zn requirement and root-to-shoot translocation rates in hyperaccumulators. Overexpression of the AhHMA4 gene in Arabidopsis induced typical markers of Zn deficiency. However, the degree of overexpression measured in hyperaccumulators is in general far beyond that induced by Zn deficiency in nonhyperaccumulator species. We currently have very little knowledge about the way plants gauge their metal status. Metal-sensor proteins have been demonstrated in bacteria (Busenlehner et al., 2003) and yeast (Rutherford & Bird, 2004) but have not been established in plants. The existence of metal sensors in plants does suggest an interesting line of future study.

Another major issue of gene regulation is the tissue specificity of gene expression. Although microarray studies have provided new insights, the tissue specificity of genes has not been addressed yet. Metals in hyperaccumulators are unevenly distributed over foliar tissues, and there may be differences not only between hyperaccumulator species but even between populations in the tissue-specific accumulation of the same metal. For example, in T. caerulescens, the highest concentrations of Cd and Zn are found in the epidermal vacuoles (Ma et al., 2005), but in A. halleri the highest concentrations are found in mesophyll cells (Küpper et al., 1999, 2000). Thus, global transcriptomic analysis at the organ level might obscure useful information. For most of the genes, neither the tissue-specific expression patterns nor the cellular localization of their products are known. This greatly limits the prediction of their role in hyperaccumulation. Future advances in the molecular study of hyperaccumulation will require tissue-specific information at both the single gene and global levels. The development of a cell sorting system is presently hampered by technical bottlenecks such as low transformation efficiency. The possibility of transforming specific cell types of hyperaccumulators may help to decipher tissue specific adaptations. In the future, combination of cell sorting with microarray analysis will undoubtedly identify new players in hyperaccumulation mechanisms and provide novel genes for bioengineering of phytoextraction or biofortification.

The complexity of hyperaccumulation is far from being understood not only at the tissue level, but also at the subcellular level. Metal transport has been studied mainly at the plasma membrane or at the tonoplast. It is expected, however, that general metal homeostasis, including processes at all other endomembranes, will be modified in hyperaccumulators. Furthermore, transport mechanisms are still poorly understood for most of the metal transporters highlighted in this review. The form of the metal that is handled by transporters is also a matter of debate and the possible role of metal-chaperones has so far only been demonstrated for Cu.

In general, there is a lack of studies at the protein level (in particular of membrane proteins) in hyperaccumulators. In the absence of protein data or functional study, the biological significance of changes at transcript levels remains to be established.

Despite important recent progress enabled by the application of new techniques, only a tiny piece of the whole hyperaccumulation puzzle has been uncovered. Plant scientists’ skills are urgently required to unlock the remaining secrets of hyperaccumulators.

Acknowledgements

The authors thank the Belgian Science Policy (project PAIVI/33) and the Fonds National de la Recherche Scientifique (FRFC project 2.4.583.08.F.) for financial support, and the COST839 network for providing excellent opportunities of discussion.

Ancillary