Natural genetic variation in caesium (Cs) accumulation by Arabidopsis thaliana

Authors

  • Katharine A. Payne,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    2. Plant Sciences Division, School of Biosciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK;
    Search for more papers by this author
  • Helen C. Bowen,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    Search for more papers by this author
  • John P. Hammond,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    2. Plant Sciences Division, School of Biosciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK;
    Search for more papers by this author
  • Corrina R. Hampton,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    2. School of Biosciences, University of Birmingham, Birmingham, B15 2TT, UK
    Search for more papers by this author
  • James R. Lynn,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    Search for more papers by this author
  • Andrew Mead,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    Search for more papers by this author
  • Kamal Swarup,

    1. Plant Sciences Division, School of Biosciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK;
    Search for more papers by this author
  • Malcolm J. Bennett,

    1. Plant Sciences Division, School of Biosciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK;
    Search for more papers by this author
  • Philip J. White,

    1. Horticulture Research International (HRI), Wellesbourne, Warwick, CV35 9EF, UK;
    Search for more papers by this author
  • Martin R. Broadley

    Corresponding author
    1. Plant Sciences Division, School of Biosciences, University of Nottingham, Sutton Bonington Campus, Loughborough, LE12 5RD, UK;
      Author for correspondence: Martin R. Broadley Tel: +44 (0)115 9516382 Fax: +44 (0)115 9516334 Email: martin.broadley@nottingham.ac.uk
    Search for more papers by this author

Author for correspondence: Martin R. Broadley Tel: +44 (0)115 9516382 Fax: +44 (0)115 9516334 Email: martin.broadley@nottingham.ac.uk

Summary

  • • Ingestion of caesium (Cs) radioisotopes poses a health risk to humans. Crop varieties that accumulate less Cs in their edible tissues may provide a useful countermeasure. This study was performed to determine whether quantitative genetics on a model plant (Arabidopsis thaliana) might inform such ‘safe’-crop strategies.
  • • Arabidopsis accessions and recombinant inbred lines (RILs), from Landsberg erecta (Ler) × Cape Verdi Island (Cvi), Ler × Columbia (Col), and Niederzenz (Nd) × Col mapping populations, were grown on agar supplemented with subtoxic levels of Cs.
  • • Shoot Cs concentration varied up to three-fold, and shoot f. wt varied up to 25-fold within populations. The heritability of growth and Cs accumulation traits ranged from 0.06 to 0.28. Four quantitative trait loci (QTL) accounted for > 80% of the genetic contribution to the total phenotypic variation in shoot Cs concentration in the Ler × Col population.
  • • QTL identified in this study, in particular, QTL co-localizing to the top and bottom regions of Chromosomes I and V in two different mapping populations, are amenable to positional cloning and, through collinearity, may inform selection or breeding strategies for the development of ‘safe’ crops.

Introduction

Radioactive 137Cs (physical half-life = 30.2 years) is released during nuclear weapons manufacture, nuclear power production and nuclear fuel cycle operations (White & Broadley, 2000). The deposition of 137Cs to soils, and its subsequent transfer to the food chain via the soil-to-plant pathway, can constitute a major radiological hazard (Gillett et al., 2001). For example, four million people in Belarus, Russia and Ukraine currently live in areas where 137Cs deposition densities to soils exceed activities of 37 kBq m−2 following the Chernobyl accident in 1986. The plant-availability of radiocaesium in these soils remains high, despite the tendency for radiocaesium to become ‘fixed’ to clay minerals in soils (Smith et al., 2000). In the Ukraine, > 70 000 ha of land were still excluded from agricultural use in 1999 because of high levels of radiocaesium, in addition to the 30 km Exclusion Zone surrounding the nuclear power plant itself. Agricultural countermeasures to minimize 137Cs in the food chain are employed within areas of the former Soviet Union. However, the dose received from food by some rural populations still exceeds the recommended maximum level of 1 mSv year−1 (Beresford et al., 2001). Since there are differences between and within plant species in their ability to accumulate Cs in their shoot tissues (Broadley et al., 1999; White et al., 2003), it may be feasible to exploit this variation and select and/or breed ‘safe’ crops, that accumulate less 137Cs in their shoot tissues, as a countermeasure to reduce ingested radiological doses to humans.

Caesium entry to plants appears to be mediated by specific cation transporters, including voltage-insensitive cation (VIC) channels and high-affinity potassium (KUP/HAK) transporters (Rubio et al., 2000; White & Broadley, 2000; Broadley et al., 2001; White & Davenport, 2002). Caesium loading to the xylem is likely to be mediated by stelar outward-rectifying K channels (SKOR; Gaymard et al., 1998). Since specific cation transporters may be important determinants of shoot Cs accumulation, altering their expression could be a strategy for developing safe crops (White et al., 2003). However, numerous candidate genes, within several different large gene families, are implicated in Cs transport in Arabidopsis (Mäser et al. 2001; White et al., 2003). Thus, gene redundancy or functional compensation (Hirschi, 2003) may limit the application of functional approaches to resolving mechanisms controlling Cs accumulation by plants.

Shoot Cs accumulation is likely to be controlled by numerous plant genes and behave as a complex quantitative trait. To dissect complex, quantitative traits, it is possible to exploit natural genetic variation within a species and use molecular-based quantitative trait loci (QTL) analyses to identify candidate genetic loci and genes that impact on the trait of interest (Alonso-Blanco & Koornneef, 2000; Paran & Zamir, 2003). This information can subsequently be used to direct crop selection and/or improvement strategies (King, 2002). Molecular-based QTL analyses in plants involve screening the progeny of a cross between two parental accessions whose chromosomes have been mapped with molecular markers that are polymorphic between accessions. By mapping the phenotypes of the progeny to the pedigree of each marker, information on the genetic and environmental effects on the trait, the heritability of the trait, and the genetic loci (i.e. QTL) controlling the trait can be determined simultaneously. In Arabidopsis, several populations of immortalized progeny, or recombinant inbred lines (RILs), are available for such studies. These include the Landsberg erecta (Ler) × Cape Verdi Island (Cvi) mapping population of 162 RILs (Alonso-Blanco et al., 1998b), a mapping population of 100 RILs from a Ler(0) × Columbia (Col-4) cross (Lister & Dean, 1993) and a population of 98 RILs from a Niederzenz (Nd-1) × Columbia (Col-3 and Col-5) cross (Deslandes et al., 1998).

Once QTL have been identified, positional cloning using, for example, near-isogenic lines (NILs), substitution lines and single-nucleotide polymorphism analyses can be integrated with functional studies of candidate genes to resolve the location of important genes that impact on the trait of interest (reviewed by Borevitz & Nordborg, 2003). Even without positional cloning, QTL identified in Arabidopsis can still be used to direct crop improvement strategies by exploiting conservation of gene order (collinearity) between Arabidopsis and a target crop species to develop closely linked markers (King, 2002). In studies of plant mineral nutrition, QTL have recently been identified in Arabidopsis that impact on plant nitrogen-use efficiency (Rauh et al., 2002; Loudet et al., 2003) and tolerance to Al toxicity (Kobayashi & Koyama, 2002; Hoekenga et al., 2003). The characteristics of shoot Cs accumulation (Cs is a nonessential plant mineral) in Arabidopsis may therefore be amenable to a similar genetic dissection. The aim of this study was therefore to quantify and dissect the natural genetic variation in Cs accumulation traits in Arabidopsis, to determine if quantitative genetics might contribute to strategies to develop safe crops with less Cs in their shoot tissue.

Materials and Methods

Three experiments were conducted to quantify and dissect the natural genetic variation in Cs accumulation traits in Arabidopsis, which had the following sequential aims: (a) to determine the appropriate experimental conditions under which to measure growth and Cs accumulation by large numbers of Arabidopsis accessions; (b) to quantify the variation in the growth and Cs accumulation characteristics of natural Arabidopsis accessions; and (c) to identify QTL impacting on the growth and Cs accumulation characteristics of three populations of Arabidopsis RILs. For all experiments, seeds were obtained from the Nottingham Arabidopsis Stock Centre (NASC, Nottingham, UK). Seeds were washed in 70% ethanol, rinsed in deionised water and surface sterilised using NaOCl (1% active chlorine). Seeds were then rinsed and imbibed for 6 days in sterile deionised water at 4°C to break dormancy. Following imbibition, six seeds from each of the accessions were sown into an unvented, polycarbonate culture box (Sigma-Aldrich, Dorset, UK). Seeds were sown on 75 ml 0.8% (w/v) agar containing 1% (w/v) sucrose and a basal salt mix at 10% of the full-strength formulation (Murashige & Skoog, 1962). The agar was amended with 133Cs and 134Cs salts, as described in the following paragraph. Boxes were placed in a growth room set to 24°C, with 16 h light d−1. A bank of 100 W 84 fluorescent tubes (Philips, Eindhoven, The Netherlands) provided a photosynthetically active (between 400 and 700 nm) photon flux density of 50–80 µmol photons m−2 s−1 at plant height.

To determine the variation in Cs accumulation by Arabidopsis, CsCl was added to the agar at 1 µm[133Cs+]ext; within the concentration range found naturally in the soil solution (White & Broadley, 2000) and below levels deleterious to the growth of Arabidopsis. This assumption was tested during a background experiment in which seeds of parents of one of the mapping populations, the Landsberg erecta (Ler, N8581) and Cape Verdi Island (Cvi, N8580) accessions were grown, under the conditions described previously, at 0.1, 0.3, 1, 3 and 10 µm[133Cs+]ext, radiolabelled with 18.33 kBq 134Cs µmol−1 133Cs. After 14, 18 and 21 d, no deleterious growth effects were detected in shoots or roots of Ler or Cvi up to [133Cs+]ext of 10 µm (Fig. 1). Thus, all subsequent experiments were conducted at 1 µm[133Cs+]ext, radiolabelled with 18.33 kBq 134Cs µmol−1 133Cs.

Figure 1.

Shoot (a, b) and root (c, d) f. wt of Arabidopsis thaliana grown for 14 d (circles), 18 d (triangles) and 21 d (diamonds) in a nutrient-replete agar supplemented with varying amounts of 133Cs. Filled symbols represent Ler (a, c); unfilled symbols represent Cvi (b, d) accessions.

Tissue Cs concentration was determined by counting 134Cs γ-emissions for 900 s per sample on an automatic well type-gamma counter (Wallac 1480 Wizard, Perkin-Elmer Life Sciences, Turku, Finland). Since biological systems do not appear to discriminate between different isotopes of Cs (White & Broadley, 2000; Tsukada et al., 2002), 133Cs was estimated directly from measured activity of 134Cs in each sample. This lack of discrimination was further observed during detailed background experiments (data not shown).

Determining assay conditions to measure the growth and Cs accumulation characteristics of Arabidopsis

Seeds of the parents of one of the mapping populations, Ler and Cvi, were grown under the conditions described previously. Twelve seeds (six seeds of Ler and Cvi) were sown into seven boxes. Boxes were randomly spaced within the growth room. The shoot and root f. wt and shoot Cs concentration of four plants, randomly harvested, of each accession were measured 7, 11, 14, 18, 21, 25 and 28 d after sowing (DAS). This experiment was repeated on three separate occasions (n = 12 for each accession).

Quantifying the variation in the growth and Cs accumulation characteristics of Arabidopsis

Fifty-six accessions, representing the worldwide distribution of Arabidopsis, were grown under the conditions described previously. Twenty-four seeds (six seeds of four unique accessions) were sown per box, i.e. 56 accessions were sown across 14 boxes. Shoot and root f. wt and shoot Cs concentration of four plants, randomly harvested, of each accession were measured 18 DAS. This experiment was repeated on three separate occasions (n = 12 for each accession). A residual maximum likelihood (REML) analysis (Thompson & Welham, 2000) was used to estimate accession means and to test for differences between accession means assuming an unbalanced design (e.g. five of the accessions failed to germinate under the growth conditions described above).

QTL impacting on the growth and Cs accumulation characteristics of Arabidopsis

To identify QTL impacting on the growth and Cs accumulation characteristics of Arabidopsis, three mapping populations of Arabidopsis RILs were grown under the conditions described previously. The first population was a Ler × Cvi mapping population of 162 RILs (Alonso-Blanco et al., 1998b). For the experiment with Ler × Cvi, a cyclical (alpha) statistical design was adopted to comprise nine separate runs. Runs 1–3, 4–6 and 7–9 each contained all of the Ler × Cvi RILs (i.e. three runs represented one complete experimental replicate). Each run contained five standard RILs, selected from the first run to span the shoot Cs concentration range, and 47–52 other Ler × Cvi RILs. Twenty-four seeds (six seeds of four unique accessions) were sown per box and 15 boxes were sown per run; gaps in this experimental design were filled with alternative ‘guard’ accessions, as appropriate. Within each run, shoot f. wt and shoot Cs concentration of four random plants of each accession were measured 18 DAS (i.e. n = 36 for standard RILs, n = 12 for all other RILs). A REML analysis was used to estimate accession means and to test for differences between accession means assuming an unbalanced design. Two further experiments were designed to screen (1) 100 RILs from a Ler (Ler-0; NW20) × Columbia (Col-4; N933) cross (Lister & Dean, 1993) and (2) 98 RILs from a Niederzenz (Nd-1; N1636) × Columbia (Col-3; N908, and Col-5; N1644) cross (Deslandes et al., 1998). Within each of these experiments, 24 seeds (six seeds of four unique accessions) were sown per box and up to 25 boxes were sown per experimental replicate; gaps in experimental designs were filled with alternative ‘guard’ accessions, as appropriate. Each experiment comprised three replicates. Within each replicate, shoot f. wt and shoot Cs concentration of four random plants of each accession were measured 18 DAS (i.e. n = 12 for all RILs). A REML analysis was used to estimate accession means and to test for differences between accession means assuming an unbalanced design.

Preliminary statistical analyses were performed using GenStat for Windows (Sixth Edition, Release 6.1.0.200, VSN International, Oxford, UK). The alpha-designed experiment was analysed using a REML analysis. Plant and accession (line) variance components were included as a random model, within a nested model of box, run and replicate variance components for the Ler × Cvi RILs. For the Ler × Col and Nd × Col RILs, a REML analysis was used without the run variance component within the random model. For each set of RILs, the heritability, defined as the accession variance component as a proportion of the sum of all variance components, of shoot f. wt, shoot Cs concentration and Cs extraction (representing the product of Cs concentration and biomass) were estimated and compared to the phenotypic variation contributed by other variance components. Accession means and standard errors for shoot f. wt, shoot Cs concentration and Cs extraction were estimated by removing the accession variance component from the random model and including it as a fixed factor.

In two of the populations, Ler × Cvi and Ler × Col, chromosomal regions impacting on shoot f. wt, shoot Cs concentration and Cs extraction, traits were mapped to the genotype of each accession using quantitative trait loci (QTL) analyses. The genotypic data for the Ler × Cvi RILs were taken as a set of 99 markers covering most of the Arabidopsis genome (Alonso-Blanco et al., 1998b). These markers spanned 482 cm, with an average distance between consecutive markers of 5 cm and the largest distance being 12 cm (Alonso-Blanco et al., 1998a). For the Ler × Col RILs, 100 markers were kindly provided by Maarten Koornneef (University of Wageningen, The Netherlands). QTL mapping was performed with the QTL Café programme (Seaton, 2000) using the Kosambi mapping function option. Marker mean values of shoot f. wt, Cs concentration and Cs extraction were calculated for each set of RILs. Marker regressions (Kearsey & Hyne, 1994) were used to test for the location and effect of one QTL on each chromosome using 5000 simulations and the regression mean square (for 1 d.f.) and significance of the fit were calculated. If, upon visual inspection, the distribution of markers indicated the presence of > 1 QTL per chromosome, and if the significance of fitting one QTL was < 0.2, the possible location and effect of two QTLs per chromosome were explored using 5000 simulations; regression mean squares were calculated for 2 d.f. The proportion of the genetic variation contributed by each QTL was calculated as the additive effect of an allele at a QTL squared, divided by heritability (the accession variance component). These QTL analyses were supplemented using the interval mapping and multiple mapping (mqm; Jansen, 1993) options of the mapqtl programme (Van Ooijen & Maliepaard, 1996) to confirm the likelihood of the presence of QTLs.

Results

Determining assay conditions to measure the growth and Cs accumulation characteristics of Arabidopsis

The shoot f. wt of Ler and Cvi accessions increased exponentially for 18 DAS (Fig. 2a). After this time, the rate of relative shoot growth declined in both accessions. Bolting and flowering was observed in individual plants of both accessions 21 DAS. Landsberg erecta had a consistently higher shoot and root f. wt than Cvi (root data not shown). Shoot Cs concentration increased between 7 and 25 DAS in both accessions, although this increase was not significant between 14 and 18 DAS (Fig. 2b). Shoot Cs concentration was consistently higher in Ler than Cvi until 25 DAS. Since Cs concentration remained approximately constant whilst relative growth was at a maximum, the contribution of non-genetic (e.g. plant-to-plant) variance components to the total phenotypic variation in shoot f. wt and shoot Cs concentration was likely to be greatest at 18 DAS. Thus, 18 DAS was selected as the time to quantify the natural genetic variation in shoot Cs concentration in Arabidopsis in subsequent experiments.

Figure 2.

(a) Shoot f. wt and (b) shoot Cs concentration of Arabidopsis thaliana grown for 28 d in nutrient-replete agar supplemented with 1 µm Cs. Closed squares, Ler accessions; open squares, Cvi accessions (mean ± sem, n = 12).

Quantifying the natural genetic variation in the growth and Cs concentration within Arabidopsis

There was natural genetic variation in shoot f. wt and Cs concentration between 51 accessions of Arabidopsis harvested 18 DAS (Fig. 3). The 51 accessions varied 8-fold in shoot f. wt (Wald statistic for accession component = 304.8, 50 d.f., χ2P < 0.001), 2-fold in shoot Cs concentration (Wald statistic = 195.8, 50 d.f., χ2P < 0.001), and 6-fold in Cs extraction (Wald statistic = 317.2, 50 d.f., χ2P < 0.001). The N2223 (Ws-1) accession had the lowest mean shoot f. wt and the N1030 (Blh-1) accession, had the lowest Cs concentration and Cs extraction. The accessions N933 (Col-4) and N901 (Ag-0) had the highest shoot f. wt and Cs concentration, respectively. Columbia-4 (Col-4) had the highest Cs extraction.

Figure 3.

Shoot f. wt (closed squares) and shoot Cs concentration (open squares) of 51 accessions of Arabidopsis thaliana grown for 18 d in nutrient-replete agar supplemented with 1 µm Cs (mean ± sem, n = 12). Inset, frequency distributions of the same data.

Variation in shoot fresh weight and Cs concentration in three populations of RILs

From the Ler × Cvi, Ler × Col and Nd × Col populations of RILs, 159, 88, and 65 accessions were harvested, respectively; remaining accessions did not germinate. Amongst Ler × Cvi RILs (Fig. 4a,b) there was 3-fold variation in shoot f. wt (Wald statistic for accession component = 647.7, 158 d.f., χ2P < 0.001), and 3-fold variation in shoot Cs concentration (Wald statistic = 679.0, 158 d.f., χ2P < 0.001). Accessions N22059 and N22102 had the lowest and highest shoot f. wt, respectively. Accessions N22146 and N22024 had the lowest and highest shoot Cs concentration, respectively. Amongst Ler × Col RILs (Fig. 4c,d), there was 10-fold variation in shoot f. wt (Wald statistic for accession component = 394.4, 87 d.f., χ2P < 0.001) and 2-fold variation in shoot Cs concentration (Wald statistic = 291.7, 87 d.f., χ2P < 0.001). Accessions N1964 and N1992 had the lowest and highest shoot f. wt, respectively. Accessions N1992 and N1990 had the lowest and highest shoot Cs concentration, respectively. Amongst Nd × Col RILs (Fig. 4e,f), there was 25-fold variation in shoot f. wt (Wald statistic for accession component = 106.0, 64 d.f., χ2P < 0.001) and 3-fold variation in shoot Cs concentration (Wald statistic = 113.4, 64 d.f., χ2P < 0.001).

Figure 4.

Frequency distributions of shoot f. wt (closed squares; a, c, e) and shoot Cs concentration (open squares; b, d, f) of (a, b) 159 Ler × Cvi recombinant inbred lines (RILs) (c, d) 88 Ler × Col RILs and (e, f) 65 Col × Nd RILs of Arabidopsis thaliana grown for 18 d in nutrient-replete agar supplemented with 1 µm Cs (mean ±sem, n = 12 or 36). Arrows depict the means of parental accessions.

The traits of shoot f. wt and shoot Cs concentration approximated a normal distribution within all three populations (Fig. 4). Transgressive segregation for shoot f. wt and shoot Cs concentration occurred in all three populations. It is noteworthy that the variation in shoot f. wt and shoot Cs concentration was at least as large within the three mapping populations of Arabidopsis RILs as across the 51 accessions. These continuous phenotypic distributions and transgressive segregation indicate that shoot f. wt and shoot Cs concentration traits are controlled by quantitative gene effects.

Within each of the three populations of RILs, the contribution of plant and accession (heritability) variance components to the total phenotypic variation, within a nested model of box, run and replicate was calculated (Table 1). The heritability of shoot f. wt ranged from 0.15 to 0.28 and the heritability of shoot Cs concentration ranged from 0.06 to 0.19 in the three populations. The heritability of shoot Cs extraction was similar in all three populations, ranging from 0.15 to 0.17. The contribution of plant-to-plant variance components to the total phenotypic variation in shoot f. wt was > 0.5 in all three populations. The contribution of plant-to-plant variance components to the total phenotypic variation in Cs concentration ranged from 0.18 to 0.46 in the three populations. The contribution of box, run and replicate variance components to phenotypic variation was greater for shoot Cs concentration than for shoot f. wt.

Table 1.  The proportion of the phenotypic variation in shoot f. wt, shoot Cs concentration and Cs extraction contributed by each of the variance component nested within the residual maximum likelihood (REML) model
Variance componentLer × CviLer × ColNd × Col
shoot f. wtShoot Cs concentrationCs extractionShoot f. wtShoot Cs concentrationCs extractionShoot f. wtShoot Cs concentrationCs extraction
Replicate0.160.130.020.090.700.410.000.070.00
Run0.020.010.00
Box0.100.590.490.040.060.040.180.280.18
Accession (heritability)0.180.080.170.280.060.160.150.190.15
Plant0.530.200.320.590.180.390.660.460.67

Markers and QTL impacting on the growth and shoot Cs concentration of Arabidopsis

For shoot f. wt within the Ler × Cvi cross, there were significant (P < 0.05) negative and positive allelic effects of Ler on Chromosome I, positive allelic effects of Ler on Chromosome II, negative allelic effects of Ler on Chromosome III and positive allelic effects of Ler on Chromosome V (Fig. 5a). For shoot Cs concentration within the Ler × Cvi cross, there were significant positive allelic effects of Ler on Chromosome I and negative allelic effects of Ler on Chromosome V (Fig. 5a). Significant QTL for shoot f. wt were identified within the Ler × Cvi cross on Chromosomes II and III, accounting for 21.5% and 16. 5%, respectively, of the genetic contribution to the phenotypic variation (Fig. 5a; Table 2). Further putative QTL for shoot f. wt within the Ler × Cvi cross were identified on Chromosome II, and in the middle of Chromosome V (Fig. 5a; Table 2). Putative QTL for shoot Cs concentration were identified within the Ler × Cvi cross at the top and bottom of Chromosome I and on Chromosome V (Fig. 5a; Table 2). Each QTL for shoot Cs concentration accounted for up to 7.7% of the genetic contribution to the phenotypic variation. Four significant or putative QTL for Cs extraction accounted for 36% of the genetic contribution to the phenotypic variation within the Ler × Cvi cross (Table 2).

Figure 5.

Figure 5.

The allelic effect of markers, homozygous for Ler, on shoot f. wt (shaded bars) and shoot Cs concentration (unfilled bars) according to their position on the five Arabidopsis chromosomes, in plants grown for 18 d in nutrient-replete agar supplemented with 1 µm Cs. Marker means were calculated from phenotypic and genotypic data derived from: (a) 159 Ler × Cvi RILs; and (b) 88 Ler × Col RILs. Significant allelic effects of Ler are indicated by asterisks (*, **, *** represent P = 0.05, P = 0.01, P = 0.001, respectively). The left-hand side of each bar represents the position of the marker used in the analyses.

Figure 5.

Figure 5.

The allelic effect of markers, homozygous for Ler, on shoot f. wt (shaded bars) and shoot Cs concentration (unfilled bars) according to their position on the five Arabidopsis chromosomes, in plants grown for 18 d in nutrient-replete agar supplemented with 1 µm Cs. Marker means were calculated from phenotypic and genotypic data derived from: (a) 159 Ler × Cvi RILs; and (b) 88 Ler × Col RILs. Significant allelic effects of Ler are indicated by asterisks (*, **, *** represent P = 0.05, P = 0.01, P = 0.001, respectively). The left-hand side of each bar represents the position of the marker used in the analyses.

Table 2.  The QTL location, the Ler allelic effect, and the percentage of the genetic variance assigned to the QTL, for shoot f. wt, shoot Cs concentration and Cs extraction in Ler × Cvi and Ler× Col mapping populations of recombinant inbred lines of Arabidopsis thaliana. QTL were simulated using marker regression analyses in QTL Café. Regression mean squares and significance of simulating one QTL per chromosome are presented for all chromosomes where P < 0.2; simulations of two QTL per chromosome are presented for selected chromosomes (c.f. Figure 5)
PopulationTraitChromosome1 QTL simulated per chromosome2 QTL simulated per chromosome
Position (cM)Ler effect% genetic varianceRegression mean square (1 d.f.)PQTL 1QTL 2
Position (cM)Ler effect% genetic variancePosition (cM)Ler effect% genetic varianceRegression mean square (2 d.f.)
Ler× CviShoot f. wt (mg)II 68.0   1.5221.5  10650.00214.00.69 4.42 70.0  1.1913.09658
III 20.0 −1.3316.5  12860.001       
V 48.0   0.76 5.4   4710.11050.00.872 7.07116.0−0.43 1.7271
Shoot Cs concentration (nmol g−1 f. wt)I120.0   0.54 7.7   1690.100 0.00.312 2.61122.0  0.53 7.4106
V 96.0 −0.40 4.2   1210.15734.00.445 5.30 90.0−0.53 7.6124
Cs extraction (pmol)I 16.0−15.11 5.41678890.082       
II 66.0  25.8915.93355390.003       
III 16.0−13.91 4.61366320.105       
V 42.0  20.6410.13496680.005       
Ler× ColShoot f. wt (mg)IV 72.0 −2.5712.6  29160.041       
Shoot Cs concentration (nmol g−1 f. wt)I 78.0   0.6130.4   1460.036 4.00.47618.69 84.0  0.53123.3108
II 36.0   0.5827.4   1490.040       
IV 68.0   0.4214.4    830.035       
V 70.0 −0.3811.7    560.089       
Cs extraction (pmol)IV 76.0−26.68 8.42834630.163       

For shoot f. wt within the Ler × Col cross, there were significant negative allelic effects of Ler on Chromosome IV and positive allelic effects of Ler on Chromosome V (Fig. 5b). For shoot Cs concentration within the Ler × Col cross, there were significant positive allelic effects of Ler on Chromosomes I, II and IV and negative allelic effects of Ler on Chromosome V (Fig. 5b). A significant QTL for shoot f. wt was identified within the Ler × Col cross on Chromosome IV, accounting for 12.6% of the genetic contribution to the phenotypic variation (Fig. 5b; Table 2). Significant QTL for shoot Cs concentration within the Ler × Col cross were identified at the top and bottom of Chromosome I, in the middle of Chromosome II, and on Chromosome IV (Fig. 5b; Table 2). A putative QTL for shoot Cs concentration within the Ler × Col cross was identified on Chromosome V. These QTL accounted for over 80% of the genetic contribution to the phenotypic variation in shoot Cs concentration within the Ler × Col cross (Table 2). The location of QTL was confirmed using interval mapping within the mapqtl programme. Putative QTL regions were analysed and refined using mqm mapping within the mapqtl programme (Van Ooijen & Maliepaard, 1996), using markers as cofactors in the analysis to reduce the genetic background noise (Jansen, 1993; Jansen & Stam, 1994). Different cofactor markers were tested around the putative QTL with the final selection of cofactors maximising the LOD score. This statistical analysis narrowed the putative QTL region to c. 10 cm (data not shown).

Discussion

There is natural genetic variation in plant growth, shoot Cs concentration, and in the amount of Cs extracted per plant in Arabidopsis thaliana. These phenomena were observed in ‘natural’ accessions, and in mapping populations of Arabidopsis recombinant inbred lines (RILs). In Landsberg erecta (Ler) × Cape Verdi Island (Cvi) and Niederzenz (Nd) × Columbia (Col) RILs, harvested at a single time-point, shoot Cs concentration varied 3-fold. In Ler × Col, shoot Cs concentration varied 2-fold. Natural genetic variation in shoot f. wt and shoot Cs extraction was greater. There was significant heritability of shoot f. wt, shoot Cs concentration and Cs extraction. Putative genetic loci impacting on shoot Cs concentration and shoot Cs extraction were identified in Ler × Cvi and Ler × Col mapping populations of RILs using a quantitative trait loci (QTL) approach. These loci, in particular, QTL for shoot Cs concentration which colocalized to Chromosomes I and V in Ler × Cvi and Ler × Col, are amenable to confirmation and resolution through positional cloning in future studies. Overall, this study: (a) demonstrates the heritability of shoot Cs concentration; (b) identifies chromosomal regions in Arabidopsis which can be resolved using fine mapping and positional cloning techniques to identify candidate genes impacting on Cs accumulation characteristics; and (c) demonstrates the feasibility of developing crops with reduced shoot Cs concentration through exploiting natural genetic variation.

Shoot Cs concentration increased with increasing plant age in Arabidopsis (Fig. 2). Two previous studies reporting Cs concentration in shoots of hydroponically-grown plants during early vegetative growth have been published. These two studies differ in their observations. In one study, the shoot Cs concentration of kale (Brassica oleracea L.) increased as a function of plant age (Weaver et al., 1981), whilst in the other, the shoot Cs concentration of wheat (Triticum aestivum L.) remained approximately constant (Smolders & Shaw, 1995). Thus, the temporal pattern of Cs accumulation in Arabidopsis shoots superficially resembles that of kale. It is not possible to explore the physiological processes that underpin this increase in shoot Cs concentration with plant age from the present study. However, this pattern of behaviour for Cs contrasts with the decline in shoot K concentration with plant age, observed in a variety of crop plants grown to maturity (Greenwood & Karpinets, 1997). The increase in shoot Cs concentration with plant age in Arabidopsis is consistent with the observations that, within the angiosperms, species of Brassicaceae have the potential to accumulate high levels of Cs in their shoot tissues, and that there is considerable variation in this trait between and within Brassicaceae species (Broadley et al., 1999; White et al., 2003).

Experimental design strongly affects the quality of QTL analyses, including the number of QTL detected, the accuracy of their map positions, and their phenotypic effects (Alonso-Blanco & Koornneef, 2000). It is therefore appropriate to maximize estimates of the contribution of non-genetic variance components to the total phenotypic variation, within the constraints of the experimental system, to ensure that the QTL identified are robust. In this study, two approaches were adopted to maximise estimates of the contribution of non-genetic variance components to the total phenotypic variation. First, a preliminary experiment to determine appropriate sampling conditions was performed using the parents of one of the mapping populations of RILs, Ler and Cvi. Since estimates of the contribution of non-genetic variance components (e.g. plant-to-plant variance components) to the total phenotypic variation in shoot Cs accumulation traits are likely to be greatest during periods of maximum plant growth, all subsequent QTL analyses were performed at 18 DAS. Second, the contributions of genetic and non-genetic variance components to the total phenotypic variation in shoot f. wt, shoot Cs concentration and Cs extraction were estimated using cyclical (alpha) experimental designs and REML analyses. This design and analysis strategy maximised the number of unique accession-to-accession comparisons whilst minimising the loss of information on the contribution of non-genetic variance components including plant-to-plant, box-to-box, and replicate-to-replicate components. Thus, estimates of the genetic contribution to the total phenotypic variation (i.e. the accession-to-accession component) were unlikely to be overestimated.

The contribution of genetic variance to the total phenotypic variation is often referred to as heritability. Heritability indicates the potential ease of conventional directed breeding efforts to deliver new crop phenotypes (Macnair, 2002) and estimates lie between one (all variation due to additive genetic factors) and zero (all variation due to environmental factors). In this study, heritability was interpreted as the contribution of genetic variance components to the total phenotypic (genetic + environmental + genetic × environmental) variation, which is also called broad-sense heritability. The heritability of shoot f. wt, shoot Cs concentration and Cs extraction ranged between 0.06 and 0.28, indicating that genetic variance components contributed up to 28% of the total phenotypic variation in Cs accumulation traits. Since heritability estimates depend on specific experimental design and analytical techniques, and since no studies on Arabidopsis mineral nutrition were identified in which all contributions of genetic and non-genetic variance components to total phenotypic variation had been quantified simultaneously, the heritability estimates in this study can not be compared directly to traits in common with growth or biomass traits reported in other studies. However, in studies that have bulked individual plants (i.e. reduced plant-to-plant variation), heritabilities of c. 0.5 have been reported for nitrogen-use efficiency traits in Arabidopsis (Loudet et al., 2003). For other quantitative traits, for example, flowering time in Arabidopsis, heritabilities between 0.07 and 0.48 have been estimated (Van Berloo & Stam, 1999). Using a similar experimental design and analysis technique to the one reported here, heritabilities of between 0.14 and 0.41 have been estimated for phosphate-use efficiency traits in Arabidopsis (JP Hammond, unpublished observations). Overall therefore quantitative genetical studies using Arabidopsis demonstrate the potential for model plant species to inform breeding strategies for improved fertilizer efficiency crop phenotypes (Loudet et al., 2003), and to develop crops with altered shoot contents of non-essential minerals such as Cs.

Within three populations of RILs, there was up to three-fold variation in shoot Cs concentration and up to 10-fold variation in shoot f. wt and Cs extraction. Mean shoot f. wt, shoot Cs concentration, and Cs extraction within three populations of RILs showed transgressive segregation around parental mean values. This observation is consistent with the hypothesis that these traits are under the control of several or more individual genes. The most detailed dissection of phenotypic variation in growth and Cs accumulation was performed in the Ler × Cvi population of RILs. Differences between replicate experiments contributed significantly to the total phenotypic variation in growth and Cs accumulation, and represented between 0.02 and 0.16 of the total phenotypic variation. Differences between runs within replicate contributed little (a maximum estimate of 0.02) to the total phenotypic variation in growth and Cs accumulation. In the Ler × Cvi population of RILs, the majority of phenotypic variation in shoot f. wt was attributed to plant-to-plant variance components (> 0.5). This is because the plants were harvested at 18 DAS when plant growth was at its maximum, and thus differences in the f. wt between individual plants would have been at their greatest. In Ler × Col and Nd × Col RILs, the partitioning of shoot f. wt variation was similar to the Ler × Cvi RILs.

In Ler × Cvi RILs, the majority of phenotypic variation in shoot Cs concentration was attributed to the box in which the plant was grown (> 0.5). Since boxes contained identical agar within each run, the position of the box in the growth room probably affected the shoot Cs concentration, possibly through a light-effect. There was less contribution of ‘box’ to total phenotypic variation in shoot Cs concentration in Ler × Col and Nd × Col RILs. In Ler × Col RILs, there was a large contribution of ‘replicate’ to total phenotypic variation in shoot Cs concentration, possibly reflecting small differences in the composition of the agar between replicates. In Nd × Col RILs, the contribution of ‘plant’ to total phenotypic variation in shoot Cs concentration dominated the analysis. These differences in the contribution of variance components to the total phenotypic variation in shoot Cs concentration between populations of RILs, imply that genetic variance components may contribute less to the total phenotypic variation than the other sources of variation, provided that all variance components are estimated simultaneously. Thus, these data illustrate that appropriate experimental design and analysis is crucial in Arabidopsis QTL experiments.

Significant differences in marker means, and associated QTL impacting on shoot f. wt, shoot Cs concentration and Cs extraction, were identified between alleles homozygous for each parent within both the Ler × Cvi and the Ler × Col populations of RILs. Overall, at least eight significant QTL were identified in two populations that impact on the growth and Cs accumulation characteristics of Arabidopsis. Although some QTL were unique to each mapping population, QTL for shoot Cs concentration occurred on Chromosomes I and V in both the Ler × Cvi and the Ler × Col populations of RILs. Since each QTL spans > 10 cm (c. 100 genes per cm), it is not yet possible to determine if these loci co-localize. Further, to identify candidate genes impacting on Cs accumulation, it will be necessary to map these loci at greater resolution. Theoretical models and physiological studies suggest that candidate genes impacting on Cs accumulation in Arabidopsis will include plasma membrane transporters expressed in root cells (Rubio et al., 2000; White & Broadley, 2000; Broadley et al., 2001) and monovalent cation transporters localized at the tonoplast (White et al., 2004). A preliminary in silico database search (http://mips.gsf.de/) revealed the presence of three transport proteins, possibly permeable to Cs+, within a genomic region of 100 000 bp (c. 25 genes) on either side of a marker where a significant allelic effect on shoot Cs concentration was observed (Fig. 5). These are a putative plasma membrane glutamate receptor channel (GLR3.1) on Chromosome II, and a putative vacuolar K+-channel (KCO4) and a proton-coupled cation transporter (CAX11) at the top of Chromosome I.

Further progress in resolving loci and identifying candidate genes can be made in Arabidopsis using, for example, heterozygous inbred families for fine-mapping (HIF; Tuinstra et al., 1997), ‘stepped aligned inbred recombinant strains’ (STAIRS; Koumproglou et al., 2002), near-isogenic lines (NILs), and single nucleotide polymorphism (SNP) analyses. For example, Swarup et al. (1999) used NILs to confirm a phenotypic effect, to map the positions of three circadian rhythm QTL, and to identify new genes that regulate the Arabidopsis circadian system. Further, positional cloning of QTL using SNP analyses led to the identification of novel alleles that impact on flowering time (El-Assal et al., 2001). It is likely that QTL impacting on major nutritional plant traits, for example, nitrogen-use efficiency (Rauh et al., 2002; Loudet et al., 2003), aluminium tolerance (Kobayashi & Koyama, 2002; Hoekenga et al., 2003), and the accumulation of nonessential plant minerals such as Cs in shoot tissues will be resolved using these techniques in the near future and will facilitate the identification of candidate genes. However, locating approximate QTL is useful even if candidate genes are not identified precisely. For example, conservation of chromosome segments between Arabidopsis and closely related plant genomes has been well documented (King, 2002). Further, there is evidence for conservation of synteny (conserved clustering of genes and unique sequences) and collinearity (conserved order of genes or markers) over different levels of divergence, including between highly diverged plant species (reviewed by King, 2002; Dominguez et al., 2003; Gebhardt et al., 2003). Thus, regions of chromosomes that impact on a trait in Arabidopsis can be used to focus breeding strategies, for example, through marker assisted breeding (MAS). However, there would clearly be a need to determine the effects of environment (and effects of genotype × environment) on any loci before embarking on such strategies.

Conclusions

This study demonstrates the potential of using quantitative genetics in a model plant species to study the accumulation of a nonessential mineral in plants. There was a significant heritable contribution to the overall phenotypic variation in Cs accumulation characteristics in Arabidopsis. Several chromosomal regions have been identified which impact on Cs accumulation, which are now amenable to confirmation and resolution using fine mapping and positional cloning techniques. This work complements ongoing functional approaches to dissect Cs transport in plants (Broadley et al., 2001; White et al., 2003) and it is feasible that future quantitative genetic approaches at higher levels of resolution will help to position important genes that impact on ion transport and accumulation. Meanwhile, increasing our knowledge of trait heritabilities, genotype × environment interactions, and QTL impacting on Cs accumulation characteristics in model species such as Arabidopsis could accelerate breeding strategies to develop safe crops with reduced shoot 137Cs concentrations. This may ultimately lead to the selection or breeding of crop varieties for cultivation to minimise the entry of 137Cs to the food chain.

Acknowledgements

KAP was supported by a UK Biotechnology and Biological Sciences Research Council (BBSRC) Plant and Microbial Sciences Committee Studentship. JPH was supported by an HRI Gordon Browning Studentship. CRH was supported by an HRI/University of Birmingham Studentship. The research in our laboratories is supported principally by the BBSRC and by the Department for Environment, Food and Rural Affairs (UK). The authors wish to thank Nick Beresford (Centre for Ecology and Hydrology, Merlewood, Cumbria, UK) and Neil Crout (University of Nottingham, UK) for their contributions to the ideas presented in this paper. A list of the markers used, and data for each accession, is available upon request from the authors.

Ancillary

Advertisement