Identifying model metal hyperaccumulating plants: germplasm analysis of 20 Brassicaceae accessions from a wide geographical area


  • Wendy Ann Peer,

    Corresponding author
    1. Center for Environmental Stress Physiology, Department of Horticulture, 625 Agriculture Mall Drive, Purdue University, West Lafayette, IN 47907 USA;
    Search for more papers by this author
  • Mehrzad Mamoudian,

    1. Center for Environmental Stress Physiology, Department of Horticulture, 625 Agriculture Mall Drive, Purdue University, West Lafayette, IN 47907 USA;
    Search for more papers by this author
  • Brett Lahner,

    1. Center for Environmental Stress Physiology, Department of Horticulture, 625 Agriculture Mall Drive, Purdue University, West Lafayette, IN 47907 USA;
    Search for more papers by this author
  • Roger D. Reeves,

    1. Institute of Fundamental Sciences–Chemistry, Massey University, Palmerston North 5301, New Zealand
    Search for more papers by this author
  • Angus S. Murphy,

    1. Center for Environmental Stress Physiology, Department of Horticulture, 625 Agriculture Mall Drive, Purdue University, West Lafayette, IN 47907 USA;
    Search for more papers by this author
  • David E. Salt

    1. Center for Environmental Stress Physiology, Department of Horticulture, 625 Agriculture Mall Drive, Purdue University, West Lafayette, IN 47907 USA;
    Search for more papers by this author

Author for correspondence: Wendy Ann Peer Fax: +1 765 494 0391 Email:


  • • Here we report on the first phase of a funded programme to select a wild relative of Arabidopsis thaliana for use in large-scale genomic strategies, including forward and reverse genetic screens for the identification of genes involved in metal hyperaccumulation.
  • • Twenty accessions of metal accumulating species of the Brassicaceae collected from Austria, France, Turkey and the USA during spring–summer 2001 were evaluated.
  • • The criteria established for selection were: hyperaccumulation of metal (Ni, Zn); compact growth habit; reasonable time to flowering; production of ≥ 1000 seeds per plant; self-fertility; a compact diploid genome; high sequence identity with A. thaliana; and ≥ 0.1% transformation efficiency with easy selection. As part of this selection process we also report, for the first time, the stable genetic transformation of various hyperaccumulator species with both the green fluorescence protein (GFP) and the bar selectable marker.
  • • We conclude that metal hyperaccumulation ability, self-fertility, seed set, transformation efficiency and a diploid genome were the most important selection criteria. Based on an overall assessment of the performance of all 20 accessions, Thlaspi caerulescens Félix de Pallières showed the most promise as a model hyperaccumulator.


Metal hyperaccumulating plants (Brooks et al., 1977) have been defined as plants that accumulate greater than 1000 µg g−1 Ni, 10 000 µg g−1 Zn or Mn, 1000 µg g−1 Co or Cu and 100 µg g−1 Cd when grown in native soils (Baker & Brooks, 1989; Reeves, 1992; Baker et al., 2000). However, it is clearly important to remember that these definitions provide only a working framework and should be applied with caution (Baker & Whiting, 2002). Reports of newly characterized metal hyperaccumulating plants have appeared with increasing frequency since 1885, when Baumann quantitated the Zn content of Thlaspi caerulescens (previously Thlaspi calaminare) at 17 000 µg g−1. Interest in metal hyperaccumulating plants increased after they were proposed as a means to phytoremediate metal polluted areas (Chaney, 1983; Baker et al., 1991, reviewed in Salt et al., 1998; Pilon-Smits & Pilon, 2002).

Despite recent advances (reviewed in Persans & Salt, 2000; Assunção et al., 2001; Clemens, 2001; Macnair, 2002; Pollard et al., 2002) the mechanisms underlying metal hyperaccumulation are still largely unknown. As such, there is a need to develop a model system for the molecular genetic study of metal hyperaccumulation in plants. With this goal in mind we have initiated a planned effort to identify a metal hyperaccumulating species for use in large-scale genomic efforts to uncover genes involved in metal hyperaccumulation. Such efforts will include the development of T-DNA mutagenized populations for use in both forward and reverse genetic screens.

Arabidopsis thaliana has become a model molecular genetic system for the study of basic plant biology due to its extensive genetic characterization, compact genome, known genomic sequence, compact growth habit, and the availability of a wide variety of tools for its molecular genetic manipulation. However, it does not accumulate metal. Fortunately, c. 25% of the documented hyperaccumulating species are, like A. thaliana, members of the Brassicaceae. Identification of an accumulating member of the Brassicaceae that is most similar to A. thaliana and therefore amenable to large-scale genome projects is therefore our first priority.

In 2001, a co-ordinated effort was initiated to collect and characterize accessions of hyperaccumulating annual/biannual members of the Brassicaceae with the objective of determining the best wild relative of A. thaliana to use as a model system to study the molecular genetics of metal hyperaccumulation. These collections were made broadly so as not to limit our final species choice. For example, known self-incompatible species were included in the collections because published information on the extent of self-incompatibility and the ability to overcome this incompatibility are limited. The 20 accessions collected in spring–summer 2001 have been evaluated based on the following criteria: does the species hyperaccumulate Zn or Ni in defined soil media? Does it have a compact growth habit amenable to greenhouse propagation? Can it flower, self-pollinate, and set sufficient seed on a reasonable time scale? Is it a diploid with a reasonably compact genome suitable for genetic studies? Can it be stably transformed with Agrobacterium tumefaciens T-DNA and can the transformants be selected easily? and is it closely related to A. thaliana?

These criteria were designed as simple, experimentally determined, qualitative parameters that taken together, allow the primary selection of plants suitable for our planned large-scale genomic efforts. Further optimization of various parameters such as time to flowering and transformation efficiencies are planned on accessions that pass this primary screening. Here we report on the primary evaluations of the 2001 collections which included Biscutella didyma L., Cochlearia aucheri Boiss., C. sempervivum Boiss. & Bal., Thlaspi caerulescens J. & C. Presl, T. goesingense Halácsy, T. oxyceras (Boiss.) Hedge, T. rosulare Boiss. & Bal., T. violascens Boiss. A. thaliana (L.) Heynh. (Col-0) is the genomic model; A. lyrata (L.) O’Kane & Al-Shehbaz (Saucon), is a nonaccumulator Arabidopsis control and T. perfoliatum L. (Vandoeuvre-lès-Nancy), is a nonaccumulator Thlaspi control which were included for comparative purposes.

Materials and Methods

Seed collection  The accessions used in this study are Biscutella didyma (Figeac, France), Cochlearia aucheri (Pülümür and Refahiye Imranlý 39°53.47′ N, 38°27.09′ E, Turkey), C. sempervivum (Elaziğ:Hazar and Pinarbaşi-Sariz, Turkey), Thlaspi caerulescens (Col du Mas de l’Aire 44°25′49 ″ N, 3°59′16″ E; Le Bleymard 44°28′17″ N, 3°43′43″ E; Les Malines 43°55′28″ N, 3°37′06″; Puy de Wolf (synonym Noccea firmiensis Meyer) 44°32′59″ N, 2°18′36″ E; St. Félix de Pallières 44°02′22″ N, 3°56′17″ E, and Viviez 44°33′33″ N, 2°13′24″ E, France), T. goesingense (Redschlag, Austria), T. oxyceras (Osmaniye 37°04.06′ N, 36°24.24′ E, Refahiye Imranlý 39°51.13′ N, 38°22.89′ E, and Zorkun yayla 36°58.143′ N, 34°20.530′ E, Turkey), T. rosulare (Içel A 36°57.861′ N, 34°24.971′ E and Içel B 36°57.253′ N, 34°24.462′ E, Turkey), sp. of T. violascens (Osmaniye 36°58.143′ N, 34°20.530′ E, Turkey). Three additional accessions were collected, but their analyses are not included here because either insufficient germination limited evaluation or the species’ identity is currently being confirmed. A. thaliana (Col-0, ABRC, Columbus, OH, USA) is the model species to which all accessions are compared. Arabidopsis lyrata (Saucon, USA) and T. perfoliatum (Vandoeuvre-lès-Nancy, France) were included as nonmetal accumulator controls. This is the first time that the majority of these plants have been germinated, grown, and induced to flower under greenhouse conditions. Seeds were collected from June until August, 2001, in France, Turkey, and the USA; seeds from Austria were collected in 1996. Locations of the populations have been previously described (Reeves & Brooks, 1983; Reeves et al., 2001a, 2001b; Peer & Murphy, 2003).

Plant growth conditions

Germination Seeds from France, Austria and the USA were sown on soil, cold-treated at 4°C for 7 d, moved to the greenhouse and germinated. Thlaspi goesingense seeds were germinated under shade cloth. Determining germination conditions has proved challenging for the accessions collected in Turkey, with optimal germination occurring when seeds were sown on soil and placed at 27°C in the greenhouse. Most accessions germinated evenly. Germination occurred within one to four weeks depending on accession.

Metal accumulation experiment Sunshine Mix #2 (J. G. Smith & Co., Batavia, IL, USA) soil was augmented with Ni(NO3)2·6H2O or Zn(NO3)2·6H2O for a final concentration of 100 µg g−1 dry weight Ni, 100 µg g−1 dry weight Zn, or 100 µg g−1 dry weight Ni and 100 µg g−1 dry weight Zn. Sunshine Mix has c. 0.2 µg g−1 d.wt Ni, c. 8.3 µg g−1 d.wt Zn and undetectable Cd before soil amendment. MetroMix 360 (The Scotts Co, Marysville, OH, USA) was used initially, but due to variation in lots, Sunshine Mix was utilized for all Ni and Zn metal accumulation data shown; 50 ng g−1 dry weight Cd was present in the MetroMix 360. Plants were grown in growth chambers (Percival, Percival Scientific, Inc., Perry, IA, USA) under 120 µmol m−2 s−1 white light with fluorescent and incandescent bulbs, 16 h d, 20°C day, 18°C night and 60% relative humidity. Plants were fertilized with 0.1× Hoagland's solution (Sigma, St. Louis, MO, USA).

Flowering In order to induce flowering in the biennial/winter annuals, plants were grown in the greenhouse (12 h light, 21°C day, 20°C night) for 10 weeks, vernalized (8 h light, 4°C) for 7, 10 or 12 weeks, then returned to long days to induce bolting (Halevy, 1985; Ferguson et al., 2001). Plants were fertilized with Peters 20-20-20 when grown under long days and received water only when grown under short days.

ICP-MS analysisc. 0.02 g samples of dry (92°C, 24 h) shoot material were digested in 2.5 ml of concentrated HNO3 for 4 h at 118°C, diluted to 16 ml, and analysed for Zn, Ni, and Cd using an ICP-MS (Inductively Coupled Plasma Mass Spectrometer). Gallium was included in each sample as an internal standard. Calibration curves were comprised of NIST-traceable standards and drift correction was applied.

Transformation and selection Plants that had bolted and had at least one opened flower were transformed by floral dipping or floral spraying with Agrobacterium tumefaciens C58 pGV3850 containing pCAMBIA3302 with the plant selectable marker bar, conferring resistance to the herbicide phosphinothrycin (Basta®) and the reporter construct mgfp5. pCAMBIA3302 was constructed by replacing the GUS reporter from pCAMBIA3301 with mgfp5 excised from pCAMBIA1302 (CAMBIA, Black Mountain ACT, Australia). Common to both floral transformation procedures, an A. tumefaciens culture was grown to mid-log phase and resuspended in 5% sucrose and 0.02% Silwet L-77 (Lehle Seeds, Round Rock, TX, USA). 0.05% Silwet L-77 was found to be too high a concentration and resulted in aborted inflorescences. In floral dipping, the inflorescence stem was dipped into the Agrobacterium solution for a few seconds; in floral spraying, the inflorescence was lightly sprayed with the Agrobacterium solution until wet, and the application was repeated 5 d later. After the floral transformation, the plants were kept in high humidity conditions under a shade cloth for 24 h, since it was observed that both precautions were necessary for flower development to continue. Selection of transformed seeds consisted of applying 50 µg ml−1 herbicide for two consecutive days at 1-week intervals for a total of 3 weeks. More frequent applications of herbicide (three consecutive days per week) resulted in fewer transformed survivors. Less frequent applications of herbicide (once per week) resulted in longer screening times. Plants were screened for GFP fluorescence before and after herbicide selection to determine sensitivity of transformed plants to the herbicide. GFP fluorescence was observed using a Leica MZFLIII microscope with a UV light source and a narrow band-pass GFP3 (ex 480/40, em 510LP) filter. Images were captured with SPOT and processed with Photoshop 5.0. The presence of GFP was confirmed by Western blot using anti-GFP polyclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA).

Flow cytometry Flow cytometry was conducted as described in Galbraith et al. (1983) with the following changes: 20 mg radish leaf (Raphanus sativus cv. Sparkle) tissue was co-chopped with 40 mg sample leaf tissue as an internal standard replacing avian erythrocytes; 5 mmβ-mercaptoethanol was added to the lysis buffer. A Coulter EPICS XL flow cytometer was used for the analysis. All reagents were from Sigma.

Chromosome number Chromosome numbers for A. lyrata were counted from root tip cells as previously described (Peev, 1975).

Sequence analysis Internal transcribed spacers (ITS1 and ITS2) and 5.8S rDNA sequences were cloned for sequencing as previously described (Koch et al., 1999). Degenerate primers (forward primer 5′-CCG TAG GTC AAC CT (GC) (GC)G(AG) (AG) (AG)G-3′, reverse primer 5′-GG (AT) (AT) AT CCC GCC TGA CCT GG-3′) (MWG Biotech Inc., High Point, NC, USA) were used in species in which the first set of primers did not produce an amplification product. GenBank accession numbers for ITS sequences used in this paper: A. thaliana, AJ232900; A. lyrata, AJ232889; C. aucheri, AF336203, AF336202; C. sempervivumAY261529; T. caerulescens, AF336188, AF336189; T. goesingenseAY261528; T. oxyceras, AF336158, AF336159; T. perfoliatum, AH010924, AH010925. Blast was used to determine per cent sequence identity.

Statistical analysis Statistical analyses of the metal accumulation data were conducted using Student's t-test and anova followed by Student-Nueman-Keuls posthoc analysis (SigmaStat, SPSS Science, Chicago, IL, USA). Samples sizes for C. sempervivum (Elaziğ), A. lyrata and A. thaliana, were n = 4; T. caerulescens accessions and T. perfoliatum, n = 5; C. aucheri accessions, C. sempervivum (Pinarbaşi), T. goesingense, T. oxyceras accessions, T. rosulare accessions and species of T. violascens, n = 7. Two replicates were analysed for all accessions, except one replicate for Turkish accessions. Accessions from calamine and serpentine soils were analysed separately.

Results and Discussion

Metal accumulation

The first step in the screening process was to quantitate the metal accumulation capacity of each accession when grown in defined soil. Since Ni and Zn have been documented to be the most robustly hyperaccumulated metals we decided to focus on these metals. Certain French populations of T. caerulescens are known to accumulate Cd. However, because this appears to be a relatively limited phenomenon phylogenetically, we chose not to include accumulation of Cd as a criterion in our selection scheme.

Figure 1 shows the Ni and Zn accumulation of plants grown in soil amended to 100 ppm Ni (Fig. 1a), 100 ppm Zn (Fig. 1b) or 100 ppm of both Ni and Zn (Fig. 1c). Among the accessions collected on serpentine soils, T. caerulescens Puy de Wolf and T. oxyceras (Osmaniye) accumulated c. 1500 ppm Ni, while C. aucheri (Pülümür), T. oxyceras (Refahiye and Zorkun) and T. rosulare (Içel A and B) accumulated c. 10× the soil concentration (Fig. 1a). Nine of 12 serpentine accessions accumulated statistically more Ni than the nonaccumulator controls (P < 0.05). Nickel accumulation was statistically significant (P < 0.05) in five accessions from calamine soils, with 1000 ppm Ni accumulation in T. caerulescens from Le Bleymard and St. Félix (Fig. 1a). Zn accumulation in Ni amended soil was > 1000 ppm in T. caerulescens from Le Bleymard, Col du Mas de l’Aire, St. Félix, and Puy de Wolf (Fig. 1a). Consistent with field data, T. caerulescens Puy de Wolf from serpentine soil accumulated high levels of Zn even when Zn soil concentrations are low (Reeves et al., 2001b). Zinc accumulation was statistically significant in T. goesingense and among the other T. caerulescens accessions compared to controls (P < 0.05). It is interesting to note that the Turkish serpentine Thlaspi species do not accumulate Zn unlike their French and Austrian relatives.

Figure 1.

Mean nickel (Ni) and zinc (Zn) accumulation quanititation. Plants of each accession were grown in soil amended with (a) 100 ppm Ni, (b)100 ppm Zn, or (b) 100 ppm Ni + 100 ppm Zn. Error bars, SD. Significance determined by anova (P < 0.001) followed by Student Newman-Keuls posthoc analysis (P < 0.05).

Accessions collected from Zn shale soils, and accumulating ≥ 3000 ppm Zn from Zn amended soil, were T. caerulescens Col du Mas de l’Aire, Les Malines, St. Félix. T. caerulescens Puy de Wolf and T. goesingense from serpentine soils also accumulated ≥ 3000 ppm Zn (Fig. 1b). None of the other accessions from serpentine soils accumulated a significant amount of Zn compared to the controls (P > 0.05) (Fig. 1b). The Ni accumulation among all accessions in Zn amended soil was very low, although some were statistically significant compared to nonaccumulator controls (P < 0.05) (Fig. 1b).

In the presence of both elevated Ni and Zn in soil (Fig. 1c) there appears to be both positive and negative synergistic effects on metal accumulation that are accession-dependent. For instance, Ni accumulation in the presence of Zn was 50% less than that in Ni soil alone in the Cochlearia species, while in the Thlaspi species from Turkey, the Ni accumulation was unaffected. In the Puy de Wolf accession of T. caerulescens, the amounts of both Ni and Zn were higher in the combined soil than in single-metal soil. Nickel accumulation was also found to be significantly higher (P < 0.05) than A. thaliana in five of eight accessions from zinc shale soils and six of 12 accessions from high Ni serpentine soils. Zinc accumulation was significant in five accessions from calamine soils and five accessions from serpentine soils (P < 0.05) (Fig. 1c).

Even though Cd accumulation in plants was not used as a selection criteria we did observe some significant differences in Cd accumulation in plants grown in 100 ppm Zn amended soil containing trace levels of Cd (50 ppb) (Fig. 2). T. caerulescens Col du Mas de l’Aire and Viviez accumulated 10 ppm Cd, significantly more than the serpentine population and the nonaccumulating controls (P < 0.01). Interestingly, elevated foliar Cd concentrations have previously be reported for T. caerulescens collected from Zn rich soils in Southern France (Robinson et al., 1998; Lombi et al., 2000; Reeves et al., 2001b). Recent work suggests that the enhanced ability to accumulate Cd in these southern France accessions is independent of the mechanism of Zn uptake (Zhao et al., 2002; Lombi et al., 2002).

Figure 2.

Mean cadmium (Cd) accumulation quantitation. Plants from Zn mine site accessions were grown in soil amended with 100 ppm Zn and containing 50 ppb Cd. Error bars, SD. n = 5. Significance determined by anova (P < 0.001) followed by Student Newman-Keuls posthoc analysis (P < 0.05).

Serpentine populations of T. caerulescens Puy de Wolf and T. goesingense have previously been observed to accumulate both Ni and Zn (Reeves & Brooks, 1983; Krämer et al., 1997; Lombi et al., 2000; Reeves et al., 2001b), and this was confirmed in our greenhouse study (Fig. 1). Interestingly, this is in contrast to the Turkish serpentine Thlaspi species T. oxyceras and T. rosulare, which do not accumulate Zn in their native habitat (Reeves, 1988) or under our greenhouse conditions (Fig. 1). Also of note, the Turkish serpentine T. violascens has been reported to accumulate elevated Zn in its native habitat (Reeves, 1988) and appeared to have a preference for Zn over Ni in our studies, similar to T. caerulescens Puy de Wolf and T. goesingense (Fig. 1). Additional elemental data from defined soil experiments and native soils are posted on the world-wide web (

Growth habit

Growth habit is clearly a continuous trait, but for the purposes of this project we have defined a compact growth habit as a plant that can grow and flower in a 4-inch pot in the greenhouse; the plant does not become root bound, rosette leaves extend ≤ 3 mm from the edge of the pot, and inflorescences are ≤ 30 mm. Root bound plants are less vigorous and more susceptible to infestation. Plants with long rosette leaves tend to shade their neighbours and become damaged when the pots are placed in close proximity, as needed to maximize greenhouse space, and damaged leaves are susceptible to disease. Plants with inflorescences > 30 mm require extra maintenance, including additional staking and rearranging fallen pots to an upright position. Metal accumulators that have a compact growth habit are C. aucheri, T. caerulescens, T. goesingense and T. violascens (Table 1). Photographs of the plants are available on the world-wide web (

Table 1.  Developmental and growth characteristics of accessions
GenusAccessionSoilLife cycleGrowth habitWeeks to floweringa% FloweringWeeks to seed ripeningYield Seeds/plantMating system
  • a

    Includes vernalization time.

  • b

    b Unvernalized.

  • c

    c nd, not determined.

Arabidopsis thalianaCol-0LoamAnnualCompact 4b100210 000Selfs
Arabidopsis lyrataSauconZn mine borderBiennialCompact16b1004     0Outcrosses
Biscutella didymaFigeacZn/Pb mineAnnualNot compact24 904     0Outcrosses
Cochlearia aucheriPülumürSerpentineBiennialCompact24 30nd     0Outcrosses
Refahiye ImranliSerpentineBiennialCompact24 50nd     0Outcrosses
Cochlearia sempervivumElaziğ: HazarSerpentineAnnualNot compactndc  0nd     0nd
Pinarbaşi-SarizSerpentineAnnualNot compact30 60nd     0Outcrosses
Thlaspi caerulescensCol du Mas de l’AireZn/Pb mineBiennialCompact24 204    60Selfs
Le BleymardZn/Pb mineBiennialCompact20 504  5000Selfs
Les MalinesZn/Pb mineBiennialCompact201004   200Selfs
Puy de WolfSerpentineBiennialCompact32 704    20Selfs
St. Félix de PallièresZn/Pb mineBiennialCompact241004  3300Selfs
ViviezZn smelterBiennialCompact201004  2000Selfs
Thlaspi goesingenseRedschlagSerpentineBiennialCompact20–241004     0Outcrosses
Thlaspi oxycerasOsmaniyeSerpentineAnnualNot compactnd  00     0nd
Refahiye ImranliSerpentineBiennialNot compact20–241000     0Outcrosses
Zorkun yaylaSerpentineBiennialNot compact20–24 500     0Outcrosses
Thlaspi rosulareIçel ASerpentineBiennialNot compactnd  0nd     0nd
Içel BSerpentineBiennialNot compactnd  0nd     0nd
Thlaspi violascensOsmaniyeSerpentineAnnualCompact20 and 241006   200Selfs
Thlaspi perfoliatumVandoeuvre-lès-NancyCalcareousAnnualCompact201002  3000Selfs

Time to flowering, seed yield and mating system

The majority of the accessions collected are winter annuals or biennials, requiring winter vernalization of the plant to flower. In order to induce winter annuals and biennials to flower on a more condensed time scale, 10-week-old plants grown under long days were transferred to short days at 4°C for a discrete numbers of weeks (see Methods section). Plants were then moved back to long days in the greenhouse and scored for weeks to flowering, and percentage of plants producing inflorescences, aborted inflorescences or no inflorescences. The treatment that produced 90–100% seed-bearing inflorescences was considered to be the best vernalization time. We expect that further optimization of time to flowering will be achievable on a smaller subset of accessions that are selected during this primary screening. Ten week vernalization time induced 90–100% seed-bearing inflorescences in T. caerulescens from Col du Mas de l’Aire, Les Malines, Viviez, T. goesingense, T. oxyceras, and T. violascens. Twelve week vernalization was required for T. caerulescens from Le Bleymard, St. Félix, and Puy de Wolf. Early and late flowerers were noted for T. violascens, T. goesingense and T. oxyceras after vernalization (Table 1).

The phenomenon of early flowering and late flowering plant has been well-defined in Thlaspi arvense, a summer and winter annual. In unvernalized plants under greenhouse conditions, early flowering plants bloom in 4–6 weeks while late flowering plants bloom in 14–21 weeks (McIntyre & Best, 1975). This difference in flowering time is due to a single gene (McIntyre & Best, 1978). Some individuals of A. lyrata, a biennial, were able to flower in 16 weeks without vernalization, and in fact lines of A. thaliana, a winter annual, have been selected in the laboratory to germinate and flower in 4 weeks without vernalization. While early and late flowering individuals of T. caeurulscens were not identified in this screen, early flowering lines may still be identified. We have observed that seeds collected at the end of the growing season for some plants (e.g. A. lytata, Thlaspi montanum, Streptanthus polygaloides) have early flowering lines (10–16 weeks), while those collected in the spring require vernalization to flower (unpublished data). Further refinement in times to flowering is possible by: increasing the critical carbohydrate content required for flowering (Levy & Dean, 1998) by growing plants under continuous light; and collecting seeds of the best candidate species at the end of summer and then selecting individuals that can germinate and flower in 16 weeks or less.

Seeds were harvested from each plant and counted to determine seed yield per plant for each treatment. Sufficient yield per plant is necessary to ensure that enough transgenic seeds can be recovered for large-scale mutagenesis projects. Assuming a transformation efficiency of 0.1% (see below) and a target population of 100 000 M1 T-DNA mutagenized lines, a seed set of 1000 would require growth and selection of c. 100–1000 trays of T2 plants (100–1000 plants per tray). T. caerulescens from Le Bleymard, St. Félix, and Viviez produce ≥ 2000 seeds per plant (Table 1). It is important to note that not all accessions grow equally well under greenhouse conditions. For example, the high seed yielding T. caerulescens Le Bleymard has a low survival rate (10%) in the greenhouse compared to T. caerulescens Viviez (100%) and St. Félix (80%). Therefore, the percentage of seed bearing plants must be included in the calculation for T-DNA mutagenesis.

Self-incompatibility is common among the Brassicaceae and was tested empirically. If isolated plants set viable seeds, then the species was considered self-compatible, if no viable seeds were collected, the plants were regarded as obligate out-crossers (Table 1). Only species that self-pollinate with high efficiency will be considered for use as a model system for large-scale mutagenesis and screening. T. caerulescens is the only metal-accumulating species in this collection that is self-fertile. Out-crossing of T. caerulescens has previously been demonstrated by Frèrot et al. (2003). Plants must be able to be out-crossed in order to do backcrosses to remove mutations not linked to the phenotype of interest in T-DNA mutagenesis projects. Backcrossing is also required for genetic complementation studies and to develop isogenic lines.

Genome size, chromosome number and ploidy

Genome size, chromosome number and polyploidy was evaluated in the accessions to identify plants with a relatively compact diploid genome and low chromosome number. Such characteristics are important for the development of an efficient insertional mutagenesis strategy. In 1997, the Angiosperm Genome Size Meeting at Kew Gardens set forth criteria for estimating the plant genome sizes via flow cytometry. This included using a plant calibration standard instead of avian erythrocytes, and specific fluorochrome and buffer recommendations. We have followed those guidelines and used Raphanus sativus as the plant calibration standard because it is a member of the Brassicaceae, its C-value is established, and its genome size is large enough to give satisfactory resolution from the sample during flow cytometry. A. thaliana was not chosen as the internal standard since its genome size was expected to be similar to the sample plants, and therefore the nuclei of the two plants would not resolve during flow cytometry. Representative flow cytometry traces of T. caerulescens Col du Mas de l’Aire with and without the R. sativus internal standard are shown (Fig. 3a,b). The intraspecific variation in genome size among the T. caerulescens accessions, for instance, may be correlated with environmental parameters involved in the adaptation of the organism (Greilhuber, 1998) (Table 2).

Figure 3.

Representative flow cymtometry traces. (a) Thlaspi caerulescens Col du Mas de l’Aire and (b) T. caerulescens Col du Mas de l’Aire with Raphanus sativus internal standard. Arrows mark 2C and 4C peaks.

Table 2.  Estimated genome sizes and chromosome numbers of selected accessions
GenusAccessionGenome size 2C (pg)Chromosome number and ploidy
Arabidopsis thalianaCol-00.342n = 2x = 10a
Arabidopsis lyrataSaucon Valley0.502n = 2x = 16
Biscutella didymaFigeac1.572n = 2x = 16b
Cochlearia aucheriRefahiye Imranli0.60nd
Cochlearia sempervivumPinarbasi-Sariz0.65nd
Thlaspi caerulescensCol du Mas de L’Aire0.682n = 2x = 14c
Le Bleymard0.702n = 2x = 14
Les Malines0.662n = 2x = 14
Puy de Wolf0.672n = 2x = 14
St. Félix de Pallières0.682n = 2x = 14
Viviez0.702n = 2x = 14
Thlaspi goesingenseRedschlag2.002n = 8x = 56d
Thlaspi oxycerasOsmaniye0.64nde
Refahiye Imranli0.67nd
Zorkun yayla0.66nd
Thlaspi rosulareIçel A0.64nd
Içel B0.64nd
Thlaspi violascensOsmaniye0.62nd

Chromosome numbers and ploidy were determined for A. lyrata during its identification, other chromosome numbers and ploidy levels were previously published (Table 2). Although a small chromosome number reduces the number of markers that need to be followed for gene linkage assessment, this criterion is not heavily weighted, whereas polyploids are not suitable for the proposed genetic studies.

How closely related are the plants to A. thaliana?

Identifying a model system in a wild species closely related to A. thaliana is desirable because genetic and technical resources that have already been developed for the A. thaliana genome projects can be applied to other genomic projects. The noncoding, intergenic transcribed spacer regions (ITS1 and ITS2) were used to establish relatedness to A. thaliana because they evolve more rapidly than coding sequences and will provide information about the evolutionary distance from A. thaliana. As expected, A. lyrata is most closely related to A. thaliana with 93% identity. All the Thlaspi species show between 87 and 88% identities, with the Cochlearia species being slightly more distantly related (84–86% identities).

Transformation and selection

For large-scale genomics projects a model species must be transformable by floral dipping or floral spraying with Agrobacterium tumefaciens, preferably have a ≥ 0.1% (see above) transformation efficiency, and transformants must also be easily selected. In this study, plants that had bolted and had at least 1 opened flower were dipped or sprayed with A. tumefaciens containing pCAMBIA3302. Two selection markers were used to confirm if stable genetic transformation had occurred, resistance to the herbicide phosphinothrycin conferred by the bar gene, and expression of the green fluorescence protein (GFP) reporter gene. Herbicide resistance rather than antibiotic resistance was chosen because seeds can be sown and selected on soil, an important advantage for large-scale T-DNA mutagenesis. Floral spraying is faster and easier than floral dipping and would be the preferred method for large-scale mutagenesis if similar transformation efficiencies are observed between the methods. Although the efficacy of floral spraying was similar to floral dipping, floral spraying resulted in a high degree of aborted inflorescences in the accessions analysed, reducing the number of recoverable transformed seeds.

All the accessions were tested for transformation, and T2 seed was recovered from A. lyrata, T. caerulescens Puy de Wolf, St. Félix, and Viviez, T. goesingense, T. perfoliatum and T. violascens. Unfortunately, attempts to transform the Cochlearia species, T. oxyceras, and T. caerulescens Le Bleymard caused all flowers to be aborted with no recoverable seed. T2 seedlings, from all species that produced seed after treatment with A. tumefaciens, were sprayed with 50 µg ml−1 phosphinothrycin twice per week, and this was found to be an effective selection regime for most of the accessions used in this study. However, seedlings of A. lyrata were found to be very sensitive to this herbicide treatment and no resistant A. lyrata seedlings were obtained. Herbicide resistant T2 seedlings were examined for GFP expression using fluorescence imaging and Western blot analysis. Of the 50–300 T2 seedlings screened from each accession we obtained between 2 and 20 individuals that were both herbicide resistant and express GFP (Fig. 4). This is this first time, to our knowledge, that any of these species have been reported to be stably transformed. This success should open up numerous new possibilities for the study of metal hyperaccumulation, including the topic of this study, T-DNA insertional mutagenesis, as well as candidate gene overexpression and knockdown using RNA interference. The initial goal of these genetic transformation experiments was to determine if the broad collection of plants under study could indeed be stably transformed, since difficulties have been encountered in other Arabidopsis species (Tague, 2001). With this initial goal achieved we are now optimizing the transformation process to establish maximal transformation efficiencies on a reduced number of target accessions.

Figure 4.

GFP fluorescence analysis of representative T2 seedlings transformed with pCAMBIA3302 by floral dipping. (a) Autofluorescence of nonherbicide-selected, nontransformed Thlaspi caerulescens Viviez seedling. (b) GFP fluorescence of herbicide-selected, transformed T. caerulescens Viviez seedling. (c) Autofluorescence of nonherbicide-selected, nontransformed T. goesingense seedling. (d) GFP fluorescence of herbicide-selected, transformed T. goesingense seedling.


The eight criteria used in this primary selection scheme are all important in identifying model hyperaccumulator plants useful for forward and reverse genetic screens using T-DNA insertional mutagenesis. However, it is clear that certain criteria will be weighted much more heavily when used as the basis for making decisions on the suitability of any particular accession as a potential model system. In our opinion the most important criteria for selection are metal hyperaccumulation ability, self-fertility, seed set, transformation efficiency and a diploid genome. Poor performance in any of these areas would immediately exclude a plant from being identified as a potential model system. Other characteristics such as time to flowering and growth habitat, though important, provide more logistical challenges that on their own are not necessarily insurmountable.

Based on an overall assessment of the performance of all 20 accessions from the 2001 collections T. caerulescens Félix de Pallières showed the most promise as a model hyperaccumulator. This accession had one of the highest foliar Zn concentrations, is self fertile with a compact diploid genome, produced 3300 seeds per plant, and is transformable. We have also successfully cloned several members of the CDF family from T. caerulescens (data not shown) using primers based on A. thaliana sequences, suggesting that 87% identity of noncoding regions is sufficient to use A. thaliana genome information to enhance any potential genomic strategy in T. caerulescens.

Once the 30 new accessions in the 2002 collections have been screened, and a top candidate identified, we plan to perform further optimization experiments with both T. caerulescens Félix de Pallières and the best candidate from the 2002 collections. Based on this information a final choice will be made as to which accession will be used for the large-scale T-DNA insertional mutagenesis and screening.


The work was supported by a grant from the National Science Foundation (0129747-IBN) to David E. Salt. We thank Rob Eddy, horticulture greenhouse manager at Purdue University. We thank Santa Cruz Biotechnology, Inc. for the kind gift of the anti-GFP polyclonal antibody. We thank Kate Carter, Kristin McFarren, Jewel Wise, Heather Arensmann and Dennis Murphy for assistance with maintaining and harvesting plants and counting seeds.