Construction of a genetic linkage map of Thlaspi caerulescens and quantitative trait loci analysis of zinc accumulation


Author for correspondence: Ana G. L. Assunção Tel: +31 317485413 Fax: +31 317483146 Email:


  • • Zinc (Zn) hyperaccumulation seems to be a constitutive species-level trait in Thlaspi caerulescens. When compared under conditions of equal Zn availability, considerable variation in the degree of hyperaccumulation is observed among accessions originating from different soil types. This variation offers an excellent opportunity for further dissection of the genetics of this trait.
  • • A T. caerulescens intraspecific cross was made between a plant from a nonmetallicolous accession [Lellingen (LE)], characterized by relatively high Zn accumulation, and a plant from a calamine accession [La Calamine (LC)], characterized by relatively low Zn accumulation.
  • • Zinc accumulation in roots and shoots segregated in the F3 population. This population was used to construct an LE/LC amplified fragment length polymorphism (AFLP)-based genetic linkage map and to map quantitative trait loci (QTL) for Zn accumulation. Two QTL were identified for root Zn accumulation, with the trait-enhancing alleles being derived from each of the parents, and explaining 21.7 and 16.6% of the phenotypic variation observed in the mapping population.
  • • Future development of more markers, based on Arabidopsis orthologous genes localized in the QTL regions, will allow fine-mapping and map-based cloning of the genes underlying the QTL.


The study of the mechanisms of metal homeostasis in plants is receiving increasing attention. Such knowledge can have important implications: for example, for human health, because it may help improve the nutritional quality of plants; for sustainable crop production, even on micronutrient-deficient soils; and for the future application of phytoremediation in metal-polluted soils. There has been some progress in establishing the molecular basis of metal homeostasis in plants, including the identification of key components (metal transporters and metal chelators) involved in metal uptake, trafficking and sequestration (Clemens, 2001; Mäser et al., 2001; Cobbett & Goldsbrough, 2002). Although most progress is being made in Arabidopsis, the study of metal hyperaccumulators (Brooks et al., 1977; Reeves, 1992), which are characterized by greatly enhanced rates of metal uptake, accumulation and tolerance (Lasat et al., 1996; Shen et al., 1997), can be of great help in unraveling the ways in which plants deal with heavy metals. Eventually this will contribute to a full understanding of the determinants of plant metal accumulation, which is at the moment still ‘a long way ahead’ (Clemens et al., 2002).

Thlaspi caerulescens is a zinc (Zn)/cadmium (Cd)/nickel (Ni) hyperaccumulator species, previously suggested to be a good model species in which to study the mechanisms of heavy metal hyperaccumulation (Assunção et al., 2003a). An important characteristic of T. caerulescens is its natural variation in important traits such as metal accumulation, metal root-to-shoot transport and metal tolerance. Comparison of accessions from different geographical and ecological environments showed a pronounced intraspecific variation for these traits (Meerts & Van Isacker, 1997; Escarréet al., 2000; Schat et al., 2000; Assunção et al., 2003b; Roosens et al., 2003). In general, this variation is of a quantitative nature, probably as a result of the effect of allelic variation at several loci (multigenic), combined with an environmental effect on each locus. This leads to a continuous phenotypic distribution of the trait in a segregating population. A continuous distribution of Zn and Cd accumulation was indeed found for segregating populations derived from T. caerulescens intraspecific crosses (Assunção et al., 2003c; Zha et al., 2004). Such quantitative genetic variation can be exploited to detect and locate the loci contributing to the Zn, Ni or Cd hyperaccumulation or tolerance traits using a so-called quantitative trait loci (QTL) analysis (Alonso-Blanco & Koornneef, 2000).

Thlaspi caerulescens belongs to the Brassicaceae family and shares 88% DNA identity in coding regions with Arabidopsis thaliana (Peer et al. 2003; D. Rigola & M. G. M. Aarts, unpublished results). This close relationship is of importance, as Arabidopsis is a model plant species with a fully sequenced and well-studied genome (AGI, 2000). Comparative genome mapping experiments can highlight the extent to which local gene order, orientation and spacing are conserved between species (Schmidt, 2000). Comparative genetic mapping experiments (for a review see Schmidt et al., 2001) have already revealed extensive conservation of genome organization (colinearity) for species of the Brassicaceae family, both at the macrosynteny and at the microsynteny levels (Kowalski et al., 1994; Cavell et al., 1998; Koch et al., 1999; Acarkan et al., 2000; Lan et al., 2000). This means that the positional information from the Arabidopsis genome can be used as an efficient tool for transferring information and resources to related plant species (Schmidt, 2000) such as T. caerulescens. Ultimately the exploitation of genome colinearity could aid the fine-mapping and subsequent map-based cloning of the genetically identified QTL (Alonso-Blanco & Koornneef, 2000; Borevitz & Chory, 2004) in T. caerulescens.

The aim of the present work was to assemble a genetic linkage map of T. caerulescens based on molecular markers and to map QTL for Zn accumulation. To this end, we used an F3 population derived from a cross between plants of the T. caerulescens accessions Lellingen (LE) and La Calamine (LC). This cross segregates for Zn accumulation, as described in Assunção et al. (2003c). The parent accessions originate from a nonmetalliferous (LE) and a calamine (LC) soil and they have been previously characterized with regard to tolerance, uptake and translocation of Zn, Cd and Ni in hydroponic culture (Assunção et al., 2003b). With respect to Zn, although they are both Zn hyperaccumulators, the LE accession is characterized by a significantly higher Zn accumulation than the LC accession, both in roots and shoots, when compared at the same level of Zn exposure (Assunção et al., 2003b). Additionally, the LC accession, originating from a calamine soil, has been shown to be much more tolerant to Zn than the LE accession, which originates from a nonmetalliferous soil (Assunção et al., 2003b). The F3 population has been genotyped using amplified fragment length polymorphism (AFLP) markers (Vos et al., 1995) to construct an AFLP-based linkage map. Additionally, PCR-based codominant markers, cleaved amplified polymorphic sequences (CAPS) and insertion/deletions (Indels) were developed for the two T. caerulescens accessions (LE and LC). These codominant markers have been used to integrate the parental genetic maps based on AFLP markers. Finally, the genetic linkage map and the root and shoot Zn accumulation phenotypes of the F3 mapping population have been used to map QTL for Zn accumulation.

Materials and methods

Plant material

A Thlaspi caerulescens J. & C. Presl F3 population was used for constructing the linkage map. The F3 mapping population consisted of 81 individuals (one individual per F3 line) derived by single-seed descent from self-fertilized F2 plants originating from a single self-fertilized F1 plant. The F1 plant was derived from a cross between a plant from the accession Lellingen (LE), originating from a nonmetalliferous site near Lellingen, Luxembourg, and a plant from the accession La Calamine (LC), originating from a strongly lead (Pb)/Cd/Zn-enriched site near La Calamine, Belgium. This cross has been previously described in Assunção et al. (2003c), in which the F3 mapping population has been referred to as F3(4).

Plant culture and phenotyping

The Zn accumulation phenotype was measured in roots and shoots of 71 individuals [71 F3(4) families] out of the 81 that constitute the F3 mapping population, and in 10–20 plants originating from the LE and LC accessions. These phenotypic data [µmol Zn g−1 root dry weight (DW) and µmol Zn g−1 shoot DW] were obtained from Assunção et al. (2003c), where the plant culture methods and Zn accumulation measurements have been described. In short, seeds were sown on moist peat and 3-wk-old seedlings were transferred to 1-l pots (one plant per pot), filled with modified half-strength Hoagland's nutrient solution, supplemented with 10 µm ZnSO4. After 3 wk, the plants were harvested and the Zn concentrations in roots and shoots were measured. Throughout the experiment the nutrient solutions were replaced twice a week and the pots were re-randomized at each solution replacement (Assunção et al., 2003c).


AFLP markers  The F3 mapping population was genotyped using AFLP marker analysis (Vos et al., 1995), which was performed as described by Qi & Lindhout (1997). Two pairs of restriction enzymes, EcoRI/MseI (E/M) and PstI/MseI (P/M), were used to generate the restriction fragments for amplification. The restriction enzymes, adaptors and primers used are listed in Table 1. Initially, 58 primer combinations (PCs), 48 E/M (+3+2) and 10 P/M (+2+3), were tested for their polymorphism rates between the parental lines LE and LC. Each of the parental lines consisted of four individuals, progeny of each of the self-fertilized parents (LE and LC). The PCs selected for use in the F3 mapping population were E32M12, E32M13, E32M14, E32M15, E32M18, E32M20, E32M22, E32M23, E35M11, E35M13, E35M17, E41M22, E45M20, E45M21, E45M22, E45M24, E45M25, P14M45, P14M47, P14M48, P14M50 and P14M52. AFLP profiles of the F1 were also obtained for all the selected PCs.

Table 1.  List of primers and adapters used to generate the amplified fragment length polymorphism (AFLP) markers
E00 (universal primer)GACTGCGTACCAATTC
EcoRI + 1 selective nucleotide E01E00 + A
EcoRI + 3 selective nucleotide E32E00 + AAC
E35E00 + ACA
E41E00 + AGG
E45E00 + ATG
P00 (universal primer)GACTGCGTACATGCAG
PstI + 0 selective nucleotide P00P00
PstI + 2 selective nucleotide P14P00 + AT
M00 (universal primer)GATGAGTCCTGAGTAA
MseI + 0 selective nucleotide M00M00
MseI + 2 selective nucleotide M11M00 + AA
M12M00 + AC
M13M00 + AG
M14M00 + AT
M15M00 + CA
M17M00 + CG
M18M00 + CT
M20M00 + GC
M21M00 + GG
M22M00 + GT
M23M00 + TA
M24M00 + TC
M25M00 + TG
MseI +3 selective nucleotide M45M00 + ATG
M47M00 + CAA
M48M00 + CAC
M50M00 + CAT
M51M00 + CCA
M52M00 + CCC

Segregating markers in the mapping population were designated according to the restriction enzymes and the primer combination used, and their sizes estimated with reference to the SequaMark 10 base ladder (Research Genetics, Huntsville, AL, USA). They were scored as dominant markers, using the specialized software package for the analysis of DNA fingerprints AFLP-Quantar® (KeyGene NV, Wageningen, the Netherlands), and designated according to the AFLP profiles of the parent lines (Fig. 1). The AFLP profile of the F1 line was used to confirm the reliability of each marker segregating in the F3 mapping population.

Figure 1.

An amplified fragment length polymorphism (AFLP) image obtained with the primer combination E32M12. It includes a set of Lellingen/La Calamine (LE/LC) F3 lines, the F1 line, the LC (calamine accession) and LE (nonmetallicolous accession) parental lines and the molecular marker (MM). Examples of different types of segregating markers are given: dominant markers, LE-specific (e32m12-192.7-LE) and LC-specific (e32m12-180.8-LC), and codominant markers, scored with AFLP-Quantar®PRO (KeyGene NV, Wageningen, the Netherlands) (e32m12-215.4-QP) and allelic band pairs (e32m12-279.1/277.2).

Codominant AFLP markers  In order to score codominant markers, the AFLP-Quantar®PRO version (KeyGene NV) was used. This program can detect codominant markers based on intensity differences of the corresponding AFLP bands. For each marker detected with the AFLP- Quantar®PRO program, the reliability was checked by comparison with the marker profiles of the parent accessions and the F1, and the codominant scoring of the mapping population was checked. For these codominant markers the letters QP were added to the marker name (Fig. 1). AFLP markers can sometimes be considered as allele markers, when bands scored as different markers actually represent different alleles of the same locus (Alonso-Blanco et al., 1998). Allelic band pairs that could be used as codominant markers were identified, taking into account the following criteria (according to Alonso-Blanco et al., 1998): the two AFLP bands should be derived from different parents, with the same primer combination; in the putative heterozygotes, and the F1, both bands should consistently show weaker intensity than in the lines containing only one band; individuals without any of the alleles do not exist. These pairs of AFLP markers were considered as single locus markers, with their names being composed of both allele names (Fig. 1). Such codominant markers have previously been identified and used in linkage map construction for other species, for example Arabidopsis (Alonso-Blanco et al., 1998), rice (Oriza nativa) (Maheswaran et al., 1997), tomato (Lycopersicon esculentum) (Saliba et al., 2000) or melon (Cucumis melo) (Perin et al., 2002).

Codominant CAPS/Indel markers  PCR-based codominant markers were developed for the accessions LE and LC. The markers were based on Expressed Sequence Tags (ESTs), from an EST library of T. caerulescens accession LC (D. Rigola & M. G. M. Aarts, unpublished results), for which homologous genes were found in Arabidopsis and which were evenly distributed over the Arabidopsis genome. Fragments of T. caerulescens genes, represented by the selected ESTs, were amplified from genomic DNA and sequenced in both accessions. An EST was generally considered the true homologue of an Arabidopsis gene if the homology search returned an E-value < e−30 and the identity was 88% or more. The results of this search were also used to locate putative intron positions in T. caerulescens, based on the assumption that their positions are conserved in Arabidopsis and T. caerulescens. For the development of a marker representing the TcZNT1 (Zn transporter) gene (Pence et al., 2000), a ZNT1 cDNA was used. Introns were preferentially targeted for amplification in T. caerulescens. When the Arabidopsis homologue did not contain an intron or the EST only represented a single exon of multiple exons in the Arabidopsis homologue (hence an intron could not be targeted for amplification), a stretch of EST sequence was targeted for amplification. Search for molecular polymorphisms between the two accessions led to the development of CAPS and Indel markers (Table 2). These PCR-based codominant markers were confirmed in both parental lines (the same individuals used to test PCs in the AFLP analysis) and in the F1, before being scored in the F3 mapping population. CAPS markers were separated on 3% standard electrophoresis grade agarose gels and Indel markers were separated on 2 or 3% Metaphor® agarose gels (Cambrex Bio Science Rockland Inc., Rockland, ME, USA).

Table 2.  List of polymerase chain reaction (PCR)-based codominant markers [cleaved amplified polymorphic sequences (CAPS)/insertion/deletions (Indels)], the Arabidopsis homologous gene, primer sequences, approximate marker fragment size (bp) and restriction enzyme used (CAPS) or allelic size difference (Indels)
CAPS/Indel markerArabidopsis homologuePrimer nucleotide sequenceFragment size (bp)Restriction enzyme (CAPS) or allelic size difference (Indels)
  1. At1g10970 (CAPS89/90) and At5g67330 (CAPS95/96) correspond to the putative zinc (Zn) transporter gene ZIP4 (Grotz et al., 1998) and the metal transporter gene NRAMP4 (Thomine et al., 2000), respectively.

  2. F, forward; R, reverse.

Indel29/30At2g36830F 5′-GTGGTAACATCACTCTCCTCCGTGG-3′35014 bp
Indel39/40At3g19820F 5′-TGTTCCTCTTTACAAGGTCGGCG-3′25020 bp
Indel47/48At5g21274F 5′-TCAGAGTTCAAGGAAGCGTTTAGCC-3′85023 bp

Map construction

A genetic linkage map of the mapping population was constructed based on dominant and codominant AFLP markers and on codominant CAPS/Indel markers. The JoinMap® 3.0 software package (Plant Research International, Wageningen, the Netherlands) (Van Ooijen & Voorrips, 2001) was used for linkage grouping and map construction. Kosambi's mapping function was applied for map-distance calculation (Kosambi, 1944).

QTL analysis

The associations between molecular markers and QTL for Zn accumulation in root and in shoot were detected using the computer program MapQTL® 5 (Kyazma BV, Wageningen, the Netherlands) (Van Ooijen, 2004). A logarithm of the odds (LOD) score of 3.0 was used as the threshold for detecting QTL (Van Ooijen, 1999). The interval mapping method and the multiple-QTL models (MQM) mapping method were used to detect and map QTL. The QTL graphs were prepared with MapChart (Voorrips, 2002).


Segregation of zinc accumulation

Zinc accumulation, in roots and in shoots, was established in hydroponically grown plants exposed to 10 µm Zn for 3 wk. For the accumulation of Zn in the roots, the LC and LE accessions exhibited different, slightly overlapping phenotypic frequency distributions, with LC showing lower Zn accumulation than LE (Fig. 2a). The frequency distributions of Zn accumulation in the shoot exhibited by LC and LE accessions did not overlap, with LC showing lower Zn accumulation than LE (Fig. 2b). Zinc accumulation in roots and in shoots was also determined in 71 individuals out of the 81 that constituted the F3 mapping population. The phenotypic frequency distribution for Zn accumulation in both roots and shoots showed segregation of these traits in the F3 population (Fig. 2c,d). All but one F3 plant had a root or shoot Zn accumulation phenotype between the lower and upper limits of the LC or LE phenotypic range. Shoot and root Zn concentrations in the F3 population were uncorrelated (r = 0.07).

Figure 2.

Frequency distribution over classes of zinc (Zn) accumulation in roots [classes correspond to 2 µmol Zn g−1 root dry weight (DW)] (a, c) and Zn accumulation in shoots (classes correspond to 5 µmol Zn g−1 shoot DW) (b, d). (a, c) Frequency distributions of Zn concentration in the roots of individuals from the nonmetallicolous [Lellingen (LE)] and calamine [La Calamine (LC)] accessions (a) and of 71 individuals out of the 81 that constitute the F3 mapping population (c). (b, d) Frequency distributions of Zn concentration in shoots of individuals from the LE and LC accessions (b) and of 71 individuals out of the 81 that constitute the F3 mapping population (d). Plants were grown for 3 wk in nutrient solution supplemented with 10 µm Zn.

Analysis and scoring of molecular markers

Fifty-eight AFLP PCs were tested in the parental lines. The sizes of fragments generated ranged from about 80 bp to 800 bp, with most fragments smaller than 500 bp. The average number of well-amplified bands per PC varied between 30 and 85, with an average of 48 bands, and the average polymorphism rate between LE and LC was 24%. Twenty-two out of the tested 58 PCs were selected for genetic mapping. On average, 15 markers were scored per PC, ranging from eight (E35M11) to 34 (E32M12). In total, of the 22 PCs, 327 segregating AFLP markers were identified and scored in the mapping population, of which 133 were LE-specific and 157 were LC-specific, 10 could be scored as codominant using AFLP-Quantar® PRO and an additional 27 allelic band pairs were identified (see Materials and methods) and scored as 27 codominant markers (Fig. 1). In the latter 27 markers, the molecular size difference between the two allelic fragments was 1 to 5 bp.

In addition to the AFLP markers, 10 PCR-based markers were developed, seven of which were CAPS and three of which were Indels (Table 2). These codominant markers were also scored in the mapping population. In total, 337 markers (327 AFLP and 10 CAPS/Indels) were scored.

The CAPS marker caps89/90 corresponds to TcZNT1, a T. caerulescens metal transporter gene, homologous to AtZIP4 (ZRT/IRT-like protein) (Grotz et al., 1998), known to mediate Zn transport and suggested to be involved in Zn hyperaccumulation in T. caerulescens (Pence et al., 2000). The CAPS marker caps95/96 is a homologue of the metal transporter AtNRAMP4 (natural resistance-associated macrophage protein) (Thomine et al., 2000). A T. caerulescens homologue, TcNRAMP4, mediates the transport of Cd, iron, manganese and Zn (R. Oomen & S. Thomine, CNRS, Gif-Sur-Yvette, France, personal communication).

Construction of an LE/LC linkage map

A genetic linkage map consisting of seven linkage groups (LGs) with in total 319 markers, including 309 AFLP (of which 37 are codominant) and 10 CAPS/Indels, was constructed. From the 337 markers scored, 333 were assembled in seven linkage groups with an LOD threshold grouping value of 5.0. During calculation of map order and distance, an additional 14 markers were omitted because they could not be reliably placed in the linkage groups. A total map distance of 449 cM was covered (Fig. 3), corresponding to an average interval of 1.6 cM between adjacent markers (absolutely linked markers were excluded in the calculation of this average). Linkage groups 1, 2, 4 and 6 have one gap of 10–15 cM and LG5 has two such intervals. LG6 covers the longest genetic distance (80.7 cM) while LG1 covers the shortest distance (44.5 cM).

Figure 3.

Lellingen/La Calamine (LE/LC) amplified fragment length polymorphism (AFLP)-based linkage map showing seven linkage groups. Cleaved amplified polymorphic sequence (CAPS)/insertion/deletion (Indel) codominant markers are shown in bold. AFLP codominant markers are shown in italic. Genetic distance between markers in cM is shown on the left of each linkage group bar. Logarithm of the odds (LOD) score values used for quantitative trait loci (QTL) analysis are indicated on the right of LG3 and LG5. The dashed line indicates a LOD value of 3. The dark bar corresponds to the area of maximum LOD value.

The great majority of the loci (89%) showed genotype ratios as expected for a segregating F3 population (0.375 : 0.25 : 0.375). For 36 markers this frequency was significantly skewed (P < 0.01), with a cluster of eight loci mapping to LG1 between positions 39 and 49 cM largely representing LE alleles, and a cluster of 22 loci mapping to LG7 between positions 37 and 53 cM with an overrepresentation of LC alleles. The other markers with a disturbed segregation were not linked to markers with distorted ratios.

Mapping QTL

Using the Zn accumulation phenotypic data, we carried out interval mapping to identify and locate QTL associated with Zn accumulation in roots and/or shoots. For Zn accumulation in roots only, we detected two QTL, one in LG3 and another in LG5 (Fig. 3). Of several markers linked to these loci, we used markers indel47/48 and e32m14-249.5/250.4 for LG3 and LG5, respectively, as cofactors for MQM mapping. Based on this analysis, we found the LG3 QTL to have a LOD value of 4.6, explaining 21.7% of the total variance, and the LG5 QTL to have a LOD value of 3.6, explaining 16.6% of the total variance. The trait-enhancing allele for the LG3 QTL originated from the LE parent and the trait-enhancing allele for the LG5 QTL originated from the LC parent. The two cofactors used for MQM analysis of both QTL are codominant markers, which allowed further assessment of the genotypes for both loci. The analysis of the genotypes of each loci indicated that the trait-enhancing allele LE3 is recessive and the trait-enhancing allele LC5 is codominant [one-way analysis of variance (ANOVA) of log-transformed data, F2,68 = 11.51 and 10.93, P < 0.001, respectively, followed by Tukey's test, to compare means]. For the genotypes of both loci, it appears that the simultaneous presence of both trait-enhancing alleles (specially LE3LE3LC5LC5 and LE3LE3LC5LE5 genotypes) has an additive effect on the root Zn accumulation phenotype (Fig. 4).

Figure 4.

Phenotypes of quantitative trait loci (QTL) genotypes. For the locus on LG3, the closely linked marker indel47/48 is used, with genotypes LE3LE3 [homozygous for the Lellingen (LE) allele], LE3LC3 (heterozygous) and LC3LC3 [homozygous for the La Calamine (LC) allele]. For the LG5 locus, the closely linked marker e32m14-249.5/250.4 is used, with genotypes LE5LE5 (homozygous for the LE allele), LE5LC5 (heterozygous) and LC5LC5 (homozygous for the LC allele). Values represent the means of the concentration of Zn in roots measured in the mapping population and are given as µmol Zn g−1 root DW. On each bar the standard error of the mean (SE) and number of plants with the corresponding genotype (n) are indicated.


Segregation of zinc accumulation

To map QTL for Zn accumulation it is necessary that this trait is segregating in the mapping population. We measured both root and shoot Zn accumulation phenotypes in individuals of the parent accessions and of the segregating F3 progeny, upon exposure to 10 µm Zn for 3 wk. These data on root and shoot Zn accumulation have previously been used to calculate the Zn accumulation, on a total plant dry weight basis, of individuals of the parent accessions and of the segregating F3 progeny [F3(4)] in Assunção et al. (2003c). The phenotypic frequency distributions of root and shoot Zn accumulation segregated in the F3 mapping population, with phenotypic frequency distributions being more or less continuous (Fig. 2c,d), suggesting polygenic or at least digenic inheritance. As we do not have the data to calculate heritability levels, it is not possible to estimate the precise numbers of loci involved in the control of these traits.

A T. caerulescens genetic linkage map

A genetic linkage map with 319 markers was generated (Fig. 3). It consists of seven linkage groups (LGs), corresponding to the haploid chromosome number of T. caerulescens. As nearly all markers could be included in the map, even at the high LOD value of 5, we conclude the seven linkage groups indeed represent the seven haploid chromosomes of T. caerulescens. With an average genetic distance between markers of 1.6 cM and with few intervals, no longer than 10–15 cM, between adjacent markers, the map covers the T. caerulescens genome very nicely. Each chromosome contains a more dense cluster of markers, probably representing the putative centromeres. Clustering of random genetic markers around the centromere, mainly as a result of centromeric suppression of recombination, has been reported in maps of several crop species such as barley (Hordeum vulgare), tomato and wheat (Triticum aestivum) (Chao et al., 1989; Tanksley et al., 1992; Qi et al., 1998). There are two clusters of skewed markers, on chromosomes 1 and 7, which show distorted segregation instead of the expected F3 segregation ratios. This happens more frequently in segregating populations [e.g. in Arabidopsis lyrata (Yogeeswaran et al., 2005) and Arabidopsis thaliana (Boivin et al., 2004)], because of the presence of alleles or combinations of alleles leading to an unfavourable phenotype. When the joint map was compared with each of the parental maps, in general a similar marker order was found (data not shown), indicating that the datasets for the parents are reliable and consistent. Taken together, all our results indicate that we have produced a robust genetic linkage map for the T. caerulescens genome.

QTL for zinc accumulation

A second goal of this research was to identify associations between molecular markers and the traits of Zn accumulation in the roots and shoots. Unfortunately, no significant QTL were found to explain the observed variation for Zn accumulation in shoots in the segregating population, although the phenotypic frequency distribution (Fig. 2d) suggests this trait to be heritable and polygenic. It is possible that the relatively small segregating population (71 individuals) does not allow the detection of significant QTL if there are many loci segregating, each with a relatively small contribution to the trait. However, for Zn accumulation in the roots two QTL were found, each with a substantial LOD score (Fig. 3). These QTL explained, respectively, 21.7 and 16.6% of the total variance and are thus of major importance. Of the two parents, the LE accession showed the highest Zn accumulation in roots and was expected to contribute the Zn accumulation-enhancing alleles. Surprisingly, however, the trait-enhancing alleles come from different parents. For the QTL on chromosome 3, the LE allele (LE3) gives rise to higher accumulation, and for the QTL on chromosome 5, it is the LC allele (LC5) that contributes to higher accumulation. The trait-enhancing allele (LE3) of the QTL on chromosome 3 seems to be recessive, suggesting that a loss-of-function mutation might be involved in the superior Zn accumulation capacity of the LE accession, relative to the LC accession. The gene responsible for this QTL could be a down-regulator of Zn uptake or, alternatively, promote Zn translocation to the shoot. It seems unlikely, however, that high root Zn concentrations would exclusively result from low rates of translocation to the shoot, as root and shoot Zn concentrations in the F3 mapping population were uncorrelated (r = 0.07) and the mean shoot Zn concentrations did not differ between the root QTL genotypes (data not shown).

The apparent additive effect on the root Zn accumulation phenotype of the trait-enhancing alleles (Fig. 4), specially with the genotypes LE3LE3LC5LC5 and LE3LE3LE5LC5, would be expected to lead to transgression in the phenotypic frequency distribution of the F3 mapping population. This was not evident; only one individual had a root Zn concentration higher than the upper limit of the analysed LE accession plants (Fig. 2c). However, the phenotypic frequency distribution of the LE accession was quite broad (Fig. 2a), probably as a result of genetic variation present in the local population from which the LE parent was taken. This is in accordance with previous findings reported by Molitor et al. (2005). The root Zn accumulation phenotype was not determined for the actual LE parental plant, as the analysis is destructive. Therefore it is not inconceivable that there are indeed F3 plants with higher root Zn accumulation than the original LE parent.

None of the detected QTL colocalized with TcZNT1 (top of chromosome 2). The QTL analysis presented here identifies loci that significantly contribute to the within-species trait variability. Although they have high variation in their Zn accumulation levels, both the LE and the LC accessions hyperaccumulate Zn when compared with a nonaccumulator species (Assunção et al., 2003b). It is conceivable that TcZNT1 is of major importance for Zn hyperaccumulation (as suggested by Pence et al., 2000 and Assunção et al., 2001), which is common to both accessions, but that there is no detectable genetic variation for this locus with respect to Zn accumulation in this intraspecific cross. This is also in accordance with the very similar mRNA expression of TcZNT1 in LE and LC, when grown for 3 wk at 10 µm Zn (Assunção et al., 2001).

The best way to further unravel the two identified QTL would be to first identify to which regions of the Arabidopsis genome the T. caerulescens QTL regions correspond and use this information to either fine-map the region further or search directly for possible candidate genes based on a presumed function in metal homeostasis. The QTL on chromosome 3 (Fig. 3) is closely linked to two T. caerulescens genes which are orthologues of Arabidopsis genes located on the top half of chromosome 5 of Arabidopsis: marker indel47/48, used as a cofactor to further analyse the QTL on LG3, and the closely linked marker CAPS65/66 are, respectively, genes At5g21274 and At5g20830. This strongly suggests conservation of genome colinearity between T. caerulescens and Arabidopsis in that region and thus that the T. caerulescens gene underlying the LG3 QTL has an Arabidopsis orthologue located on chromosome 5 in the vicinity of At5g20830/At5g21274. In this region of chromosome 5 there are three genes possibly involved in metal homeostasis, namely AtFRO4 (At5g23980) (ferric reductase oxidase, FRO) and AtFRO5 (At5g23990), both encoding ferric-chelate reductases, and AtYSL2 (At5g24380) (yellow-stripe like, YSL), encoding a yellow-striped leaf-like protein. AtFRO4 and AtFRO5 have not been studied in much detail, but they share homology to AtFRO2, which is a root cell plasma membrane ferric reductase involved in iron uptake (Robinson et al., 1999). It seems unlikely, however, that these genes would be directly involved in Zn accumulation. AtYSL2 encodes a protein demonstrated to facilitate transport of nicotianamine-chelated iron and copper. In yeast this protein was not found to complement a Zn uptake mutant (DiDonato et al., 2004).

For the QTL on chromosome 5 of T. caerulescens (Fig. 3), there is no closely linked marker corresponding to a region of the Arabidopsis genome. The only markers linking the two genomes are indel29/30 and CAPS85/86, both residing on the lower half of chromosome 2 of Arabidopsis at At2g36830 and At2g36540, respectively. These markers are around 20 cM distant from the most likely QTL position. As there is no additional information on colinearity between Arabidopsis and T. caerulescens or any other closely related species, this may well correspond to a region on another chromosome of Arabidopsis.

In this study we have constructed a reliable and robust genetic map for a segregating F3 population of a so far genetically uncharacterized plant species. Such a map is very useful for the further analysis of natural genetic variation for Zn accumulation found among wild accessions of T. caerulescens. Although we used only a limited set of markers corresponding to orthologous genes in both T. caerulescens and Arabidopsis, it was possible to make realistic suggestions about which region of the Arabidopsis genome could harbour an orthologue of the T. caerulescens gene underlying one of the identified QTL. This is not conclusive evidence that there is indeed such an orthologue, as T. caerulescens contains genes not found to have an Arabidopsis orthologue (D. Rigola & M. G. M. Aarts, unpublished results). However, this is a suitable starting point to focus on for further fine-mapping of the region in order to limit the number of putative QTL candidate genes. In the future, fine-mapping of the QTL regions in T. caerulescens and local saturation of the T. caerulescens map with more Arabidopsis ‘anchors’ will provide the required information that can eventually lead to the cloning of the QTL genes.


We thank Diana Rigola for providing Thlaspi caerulescens EST data, Petra van den Berg, Fien Meijer and Mattijs Bliek for technical advice and Prof. Dr Maarten Koornneef for critical reading of the manuscript. This research was supported by the EU project PHYTAC, contract no. QLRT-2001-00429 (to AGLA).