This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 142/3/1127    most recent pp.106.082073v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (18) Citing Articles via Google Scholar Google Scholar Articles by van de Mortel, J. E. Articles by Aarts, M. G.M. Search for Related Content PubMed PubMed Citation Articles by van de Mortel, J. E. Articles by Aarts, M. G.M. Agricola Articles by van de Mortel, J. E. Articles by Aarts, M. G.M.

GENETICS, GENOMICS, AND MOLECULAR EVOLUTION

Large Expression Differences in Genes for Iron and Zinc Homeostasis, Stress Response, and Lignin Biosynthesis Distinguish Roots of Arabidopsis thaliana and the Related Metal Hyperaccumulator Thlaspi caerulescens^1,[W]

Laboratory of Genetics, Wageningen University, 6703 BD Wageningen, The Netherlands (J.E.v.d.M., L.A.V., M.K., M.G.M.A.); Institute of Ecological Sciences, Faculty of Earth and Life Sciences, Vrije Universiteit, 1081 HV Amsterdam, The Netherlands (H.S.); ServiceXS BV, 2333 AL Leiden, The Netherlands (J.K.); Agilent Technologies, Little Falls Site, Wilmington, Delaware 19808–1644 (S.C.); and Bioinformatics Laboratory, Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, University of Amsterdam, 1100 DD Amsterdam, The Netherlands (P.D.M., E.V.L.v.T.)

	ABSTRACT

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 
The micronutrient zinc has an essential role in physiological and metabolic processes in plants as a cofactor or structural element in 300 catalytic and noncatalytic proteins, but it is very toxic when available in elevated amounts. Plants tightly regulate their internal zinc concentrations in a process called zinc homeostasis. The exceptional zinc hyperaccumulator species Thlaspi caerulescens can accumulate up to 3% of zinc, but also high amounts of nickel and cadmium, without any sign of toxicity. This should have drastic effects on the zinc homeostasis mechanism. We examined in detail the transcription profiles of roots of Arabidopsis thaliana and T. caerulescens plants grown under deficient, sufficient, and excess supply of zinc. A total of 608 zinc-responsive genes with at least a 3-fold difference in expression level were detected in A. thaliana and 352 in T. caerulescens in response to changes in zinc supply. Only 14% of these genes were also zinc responsive in A. thaliana. When comparing A. thaliana with T. caerulescens at each zinc exposure, more than 2,200 genes were significantly differentially expressed (5-fold and false discovery rate < 0.05). While a large fraction of these genes are of yet unknown function, many genes with a different expression between A. thaliana and T. caerulescens appear to function in metal homeostasis, in abiotic stress response, and in lignin biosynthesis. The high expression of lignin biosynthesis genes corresponds to the deposition of lignin in the endodermis, of which there are two layers in T. caerulescens roots and only one in A. thaliana.

Although the zinc homeostasis mechanism is supposed to be universal within plants, there are species that can accumulate large amounts of zinc without any sign of toxicity. Species accumulating more than 10,000 µg zinc g^–1 dry weight (DW; 1% [w/w]) are called zinc hyperaccumulators (Baker and Brooks, 1989). As a comparison, most plants contain between 30 and 100 µg zinc g^–1 DW and concentrations above 300 µg zinc g^–1 DW are generally toxic (Marschner, 1995). More than 400 metal hyperaccumulator species from a wide range of unrelated families have been described. About 15 of these are zinc hyperaccumulators (Baker et al., 1992; Brooks, 1994). They are mainly, though not exclusively, found to grow on calamine soils contaminated with lead, zinc, or cadmium (Meerts and Van Isacker, 1997; Schat et al., 2000; Bert et al., 2002). Thlaspi caerulescens J. & C. Presl (Brassicaceae) is one of these natural zinc hyperaccumulator species. In addition to zinc, it can also hyperaccumulate cadmium and nickel. It is a self-compatible species, showing variable rates of outcrossing in nature. T. caerulescens is closely related to Arabidopsis thaliana L. Heynh., with on average 88.5% DNA identity in coding regions (Rigola et al., 2006) and 87% DNA identity in the intergenic transcribed spacer regions (Peer et al., 2003). As in most metal hyperaccumulators, the zinc concentration in shoot tissue of T. caerulescens is often higher than in root tissue (Lasat et al., 1996; Shen et al., 1997; Schat et al., 2000).

The complex network of homeostatic mechanisms that evolved in plants to control the uptake, accumulation, trafficking, and detoxification of metals (Clemens, 2001) also applies for metal hyperaccumulators. In general, this network involves three major components: transport, chelation, and sequestration. While the physiology of metal hyperaccumulation is already understood fairly well (Clemens et al., 2002), the underlying molecular genetics is still not explored in full detail. Previously published transcript-profiling studies on copper, zinc, and iron deficiency in A. thaliana (Wintz et al., 2003) and comparative analysis of A. thaliana with the zinc- and cadmium-hyperaccumulating Arabidopsis halleri (Becher et al., 2004; Weber et al., 2004) using first generation Affymetrix chips representing a subset of only 8,300 of the approximately 30,000 Arabidopsis genes already identified several genes to respond to zinc deficiency in A. thaliana. These analyses also revealed that the transcriptional regulation of many genes is strikingly different in A. halleri compared to A. thaliana.

In this report, we describe the analysis of three transcript-profiling experiments, with the main aim of establishing which genes are most likely to be relevant for adaptation to high zinc exposure in T. caerulescens. Therefore, we examined not only the response of roots of both plant species to zinc deficiency, but also to excess of zinc. We used an Agilent whole-transcriptome, 60-mer oligo DNA microarray representing all annotated genes for A. thaliana (further referred to as Arabidopsis) and some 10,000 nonannotated genomic regions with known transcriptional activity, thus covering nearly the complete Arabidopsis transcriptome. In the intraspecific comparison, we identified the Arabidopsis and T. caerulescens genes that are differentially expressed a week after transferring the plants to low or high zinc supply. These are relevant in determining differences in the zinc homeostasis network between the two species. We also compared the differences in transcription between the two species at zinc deficiency, sufficiency, and excess supply conditions to identify the genes that are significantly more highly expressed in the hyperaccumulating species compared to Arabidopsis. We finally examined all analyses to identify any particular processes, biochemical pathways, or gene classes that could play a particular role in the adaptation to high zinc accumulation.

	RESULTS

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 

Experimental Design

To analyze the response of Arabidopsis and T. caerulescens to different zinc exposures, we aimed to compare the transcript profile of plants grown under sufficient zinc supply with plants grown under zinc deficient conditions and excess zinc conditions. To minimize variation in the bioavailability of zinc or other micronutrients, we used a hydroponic rather than soil-based culturing system. For both excess and deficient zinc conditions, the induction of severe stress to the plants was avoided by exposing them for only 1 week to these conditions. Arabidopsis plants (accession Columbia) were established to grow on a nutrient solution containing 2 µM ZnSO₄, which is sufficient to yield healthy and robust plants with normal seed set even after prolonged cultivation. After 3 weeks, the plants were transferred to fresh solutions for exposure to zinc deficiency (0 µM ZnSO₄) and excess zinc (25 µM ZnSO₄). One-third of the plants remained at sufficient zinc (2 µM ZnSO₄) as a control. From previous experiments (data not shown), we learned that plants continue to grow in zinc deficient media, while deficiency symptoms (chlorosis and necrosis) or toxicity symptoms become obvious only after about 3 weeks. Upon harvesting root tissue, the plants growing on zinc deficient and sufficient medium did not show any visible phenotypic differences, whereas plants growing on excess zinc showed a slight growth inhibition in the roots (data not shown). At this stage, plants were not flowering yet.

For T. caerulescens, a similar approach was taken. However, to properly compare the results between the two species, we aimed at maintaining comparable physiological conditions. T. caerulescens accession La Calamine is zinc tolerant as well as zinc hyperaccumulating and requires more zinc than Arabidopsis for normal growth. Therefore, a hydroponic solution containing 100 rather than 2 µM ZnSO₄ was used to grow seedlings under zinc sufficient conditions. To avoid any problems with possible precipitation of zinc or other minerals, 1 mM ZnSO₄ was used as the excess zinc exposure concentration, although we learned from previous experiments that T. caerulescens La Calamine is well able to withstand this exposure for several weeks. When root tissues were harvested after 1-week exposure, the T. caerulescens plants showed no altered phenotype that could be attributed to exposure to excess zinc or zinc deficiency.

Mineral Content in Arabidopsis and T. caerulescens

Zinc, iron, and manganese concentrations were determined in root and shoots of hydroponically grown Arabidopsis and T. caerulescens plants grown at deficient, sufficient, and excess zinc. Comparison of the metal concentration levels between these two species already displayed the typical difference between a metal hyperaccumulator and a metal nonaccumulator (Fig. 1 ). At low zinc supply (zinc deficiency), the difference is not very pronounced, and although T. caerulescens clearly contains more zinc in the leaves, the concentration in the roots for both species is between 1.2 to 1.7 µmol g^–1 DW (Fig. 1A). At sufficient zinc supply, T. caerulescens accumulates about 3 times more zinc in the roots than Arabidopsis, and the concentration in shoots is much higher (approximately 70-fold) than in Arabidopsis. At excess zinc, the zinc-exclusion strategy of Arabidopsis roots has collapsed and their zinc concentration is now about 4.5-fold higher than in T. caerulescens; this is approximately 15-fold higher compared to sufficient conditions in Arabidopsis. At this high zinc supply, Arabidopsis is still able to exclude zinc accumulation in the leaves, with a concentration about 9-fold lower than in T. caerulescens. For T. caerulescens, there is not much difference in both root and shoot concentrations of plants growing at sufficient or excess zinc.

View larger version (15K): [in this window] [in a new window]   Figure 1. Zinc (A), iron (BI: roots; BII: leaves), and manganese (C) concentrations (µmol g–1; mean ± SE) in Arabidopsis and T. caerulescens roots (white bars) and leaves (black bars). Plants were grown for 3 weeks on nutrient solution containing sufficient zinc before exposure to zinc deficiency (0 µM ZnSO4: Zn0), zinc sufficiency (2 or 100 µM ZnSO4: Zn2/Zn100), and excess zinc (25 or 1,000 µM ZnSO4: Zn25/Zn1000).

For manganese, the situation is the opposite in roots. In T. caerulescens, the manganese concentration decreases with increasing zinc supply and in Arabidopsis this only occurs upon excess zinc supply. The manganese concentration in T. caerulescens roots is about 5-fold higher under zinc deficiency than in Arabidopsis, but at sufficient zinc supply there is hardly any difference between the species. There are only few differences for the manganese concentration in leaves between the two species. Only at excess zinc, the manganese concentration in T. caerulescens decreases drastically by about 4-fold. Both T. caerulescens and Arabidopsis accumulate manganese to a higher concentration in leaves than in roots, with the exception of T. caerulescens grown under zinc deficiency.

Zinc Response in Arabidopsis

Genes responding to changes in zinc exposure conditions in Arabidopsis were identified using Agilent Arabidopsis 3 60-mer oligonucleotide microarrays containing 37,683 probes representing more than 27,000 annotated genes and more than 10,000 nonannotated genomic regions for which there is transcriptional evidence. When analyzing the data, we only considered the hybridization data of probes with P < 0.05. Only expression differences of 3-fold (between any of the three treatments) were considered to be relevant, even though lower expression differences were statistically significant at P < 0.05. According to these criteria, we identified 608 zinc-responsive genes when comparing the deficient, sufficient, and excess zinc treatments. As expected, most differences were found between the most distant conditions, zinc deficiency and excess zinc. Many genes that were differentially expressed between zinc deficiency and sufficiency or between zinc sufficiency and excess zinc were also differentially expressed between deficiency and excess zinc, whereas few genes were found to be only differentially expressed between both zinc deficiency and sufficiency, and sufficiency and excess zinc.

After hierarchical clustering (average linkage hierarchical clustering with uncentered correlation; Eisen et al., 1998) of all differentially expressed genes, four major clusters were distinguished. Cluster I (Supplemental Table S1) consists of 98 genes that are more lowly expressed under zinc deficiency compared to sufficient and excess zinc. Within this group, we find many genes with a function related to stress response and also several with metabolism-associated functions. Among the 10 genes most differentially expressed between the zinc deficiency and sufficiency exposures are three genes encoding small heat shock proteins. Other genes found in this cluster are genes encoding copper/zinc superoxide dismutases, a nodulin-like protein, a nitrate-responsive protein, an expansin-like protein, and a universal stress protein. The last one is the most highly expressed gene at sufficient zinc found in this cluster. Fifteen genes in this cluster encode proteins with an unknown function. Twenty probes correspond to transcripts that were not annotated as such in the Arabidopsis genome.

Cluster II (Table I ; Supplemental Table S2) is a large cluster consisting of 128 genes. These genes are generally more highly expressed under excess zinc conditions when compared to sufficient or deficient zinc supply. This cluster contains several metal homeostasis-related genes associated with iron rather than with zinc homeostasis. These genes encode metal transporters (IRT1, IRT2, ZIP8, MTP3, MTP8, NRAMP4, and IREG2), a nicotianamine (NA) synthase (NAS) gene (NAS1), a YS-like oligopeptide transporter (OPT3), and ferric-chelate reductases (FRO1, 2, and 3). Not only are metal homeostasis genes found in this cluster, but also some stress response genes like a disease resistance gene. In addition, metabolic genes like PAL2 (which encodes a key enzyme acting early in the phenylpropanoid biosynthesis pathway leading to flavonoids, anthocyanins, and lignins) and genes belonging to the cytochrome P450 family (CYP98A3, CYP82C2, CYP82C3, CYP82C4, CYP71B5, and CYP71B38) are found in this cluster. This cluster also contains a small set of genes encoding transcription factors of the basic helix-loop-helix (bHLH), myb, and zinc-finger families.

When comparing all three zinc exposure conditions, only genes encoding ferric-chelate reductases (FRO1 and 5), two ZIP metal transporters (ZIP3 and 9), iron superoxide dismutase (FSD1), an oxidoreductase (At3g12900), an iron-sulfur cluster assembly complex protein (At2g36260), cytochrome P450 CYP82C4, and an expressed protein (At3g59930) are differentially expressed between all conditions (Tables I and II; Supplemental Table S3).

Heterologous Microarray Hybridization

We used the same Arabidopsis array platform for heterologous hybridization with labeled T. caerulescens cDNA. From analysis of approximately 3,500 expressed sequence tags (ESTs), we previously determined that T. caerulescens shares about 85% to 90% DNA identity in coding regions with Arabidopsis (Rigola et al., 2006). However, since most probes present on the arrays were designed to fit less conserved regions of the Arabidopsis transcripts, we verified the suitability of cross-species hybridization of these arrays. First, Agilent Arabidopsis 1 oligonucleotide arrays, representing around 13,500 putative genes, were hybridized with labeled cDNA from Arabidopsis and T. caerulescens roots grown under sufficient zinc conditions. The spot intensities of the T. caerulescens hybridizations were on average only 1.7-fold lower than the spot intensities of Arabidopsis hybridizations, which is sufficient for reliable expression analysis. In addition, genomic DNA hybridization of T. caerulescens to the Agilent Arabidopsis 3 oligonucleotide array showed average 2.0-fold lower signal intensity for T. caerulescens compared to the Arabidopsis signal intensities. Overall, only probes corresponding to 220 genes did not hybridize with T. caerulescens genomic DNA (less than 3-fold less signal intensity). These 220 genes were excluded from the dataset.

Zinc Response in T. caerulescens

When comparing the expression of genes in roots of T. caerulescens plants grown on deficient, sufficient, and excess zinc media, we identified 350 genes that were significantly (false discovery rate [FDR] P < 0.05) differentially expressed (3-fold) in any of the three possible comparisons. Only 50 of these were also differentially expressed in response to different zinc exposures in Arabidopsis. Six clusters were identified in this set upon cluster analysis (average linkage hierarchical clustering with uncentered correlation). Clusters I and II (Table III ; Supplemental Table S5) consist of 38 genes that are more highly expressed at zinc deficiency compared to sufficient or excess zinc treatments. ZIP-like genes ZIP3, ZIP4, and ZIP9 are coexpressed, showing higher expression at sufficient zinc compared to excess zinc. This in contrast to the ZIP1 and ZIP2 genes, which are expressed at similar levels under zinc sufficiency and excess zinc. Other known metal homeostasis genes found in this cluster are the NAS4 and FRO5 genes. These were also found to be more highly expressed in zinc deficient Arabidopsis roots. The FSD1 iron superoxide dismutase, which was also found to be differentially expressed in Arabidopsis, is found in this cluster. The most differentially expressed gene encodes a Dirigent protein (DIR5), involved in lignin biosynthesis. This gene is hardly expressed under sufficient zinc conditions, which explains the strong differential expression.

Clusters IVA and IVB (Supplemental Table S7) consist of 19 and 14 genes, respectively, all most highly expressed under excess zinc compared to the other two conditions. Two of these genes (At5g05250 and At2g41240) are also found in a similar cluster for Arabidopsis roots (Table I; Supplemental Table S2). Compared to Arabidopsis, the Thlaspi cluster is much smaller and lacks all of the iron homeostasis genes.

The remaining 189 genes fall into two additional clusters, generally more highly expressed under sufficient than under deficient conditions (Supplemental Tables S8 and S9). Almost half of these encode proteins with an unknown function. Many of the other genes are involved in general metabolism and stress response.

Difference in Zinc Response between Arabidopsis and T. caerulescens

To identify genes that may be crucial for the adaptive differences between Arabidopsis and T. caerulescens, we compared the gene expression profiles between the two species for each of the tested physiological conditions. Taking into account that we performed a heterologous hybridization and that probes generally did not hybridize as efficiently to T. caerulescens cDNA as to Arabidopsis cDNA, we only considered significant probes with a more than 5-fold higher normalized hybridization signal in T. caerulescens compared to Arabidopsis in any of the three comparisons to be of biological relevance.

According to these criteria, in total 2,272 genes were found to be at least 5 times significantly more highly (FDR P < 0.05) expressed in T. caerulescens compared to Arabidopsis (Supplemental Table S10). Of these genes, 420 (18.5%) were not found to be expressed in roots of Arabidopsis under comparable conditions (Supplemental Table S11). A large class of 1,147 of the 2,272 differentially expressed genes has an unknown biological function. Other classes represent genes encoding proteins involved in cellular processes, transport processes, stress response, and transcription. A total of 929 genes showed little variation in expression under the three tested conditions, suggesting a constitutively higher expression in T. caerulescens roots. To test if this expression is anyhow functionally related to metal stress adaptation, the functional distribution of this group was compared to that of all 2,272 differentially expressed genes, but this did not show specific gene classes to be overrepresented or underrepresented (data not shown). Only 121 of the 2,272 genes are differentially expressed in T. caerulescens in response to different zinc exposures. The most highly expressed genes among these are the ZIP4 and IRT3 metal transporters (Tables III and IV ). Remarkable is the large difference in expression between T. caerulescens and Arabidopsis of four members of the PDF gene family (Table IV; Supplemental Table S12). These genes are especially highly expressed in T. caerulescens under deficient and excess zinc conditions compared to Arabidopsis. To facilitate the further analysis of this large class of genes, a selection was made of 235 genes, including the 50 most differentially expressed genes under zinc deficiency and genes of which the proposed function could be relevant in explaining the metal adaptation differences between both species (Table IV; Supplemental Table S12).

View larger version (81K): [in this window] [in a new window]   Figure 2. UV autofluorescence of lignin and suberin deposition in comparable cross sections of roots of hydroponically grown Arabidopsis (A, B, and C) and T. caerulescens (D, E, F, and G). A, Cross section made 2 cm from the tip of a 4-week-old Arabidopsis root showing blue UV fluorescence of the xylem vessels and the epidermis outer cell walls. B, Cross section made 2 cm from the root tip of a 6-week-old Arabidopsis root showing strong UV fluorescence of the xylem vessels and light fluorescence of the inner wall of endodermis cells. C, Cross section made 6 cm from the root tip of a 6-week-old Arabidopsis root showing strong UV fluorescence of the xylem and faint fluorescence of the inner wall of the endodermis cells. D, Cross section made 2 cm from the tip of a 4-week-old T. caerulescens root showing strong UV fluorescence of the xylem vessels and the inner walls of the endodermis cells. E, Close-up of a cross section made 1 cm from the root tip of a 4-week-old T. caerulescens root showing the UV fluorescence of the xylem vessels and the inner walls of the endodermis cells. F, Cross section made 6 cm from the root tip of a 6-week-old T. caerulescens root showing UV fluorescence of the xylem vessels and the inner or outer walls of endodermis cells. Remarkable is the apparent formation of a second layer of endodermis cells, of which the inner walls are also fluorescent. G, Cross section of a 9-week-old T. caerulescens root made 6 cm from the root tip showing two layers of endodermis, both of which show UV fluorescence. cx, Cortex; en, endodermis; x, xylem. Bar = 20 µm.

For confirmation of the microarray expression profiling data, a small subset consisting of differentially expressed genes and random genes was subjected to semiquantitative reverse transcription (RT)-PCR. In the absence of T. caerulescens DNA sequences suitable for designing species-specific PCR primers, orthologous T. caerulescens gene fragments were first amplified by low-stringency PCR using Arabidopsis-specific primers and confirmed by DNA sequencing. This sequence was used to design primers for semiquantitative RT-PCR hybridizing at comparable positions of T. caerulescens and Arabidopsis gene sequences. Expression of the target genes was studied in both root and leaf tissues of plants grown hydroponically at different zinc supply conditions (Fig. 3 ). In general, the root expression levels determined by semiquantitative RT-PCR were comparable to those determined by microarray analysis, confirming the significance of the heterologous microarray hybridization results.

View larger version (76K): [in this window] [in a new window]   Figure 3. Comparative semiquantitative RT-PCR of selected putative metal homeostasis-related differentially expressed genes (APX2 to At1g20380) and three randomly picked genes (At1g27950 to At5g50400) in Arabidopsis (At) and T. caerulescens (Tc). For amplification, species-specific primers were designed at comparable locations in each orthologous gene pair. Roots and leaves were harvested separately after 1 week of exposure of 3-week-old plants to 0, 2, and 25 µM ZnSO4 for Arabidopsis and 0, 2, 10, 100, and 1,000 µM ZnSO4 for T. caerulescens. APX2, L-ascorbate peroxidase, At3g09640; FER1, ferritin, At5g01600; FRO4, ferric chelate reductase-like, At5g23980; HAK5, potassium transporter, At4g13420; HMA4, zinc ATPase E1-E2 type, At2g19110; IRT1, Fe(II) transporter, At4g19690; NAS1, nicotianamine synthase, At5g04950; NAS2, nicotianamine synthase, At5g56080; NAS3, nicotianamine synthase, At1g09240; NAS4, nicotianamine synthase, At1g56430; ZIP1, zinc transporter, At3g12750; ZIP4, zinc transporter, At1g10970; bHLH100, bHLH transcription factor, At2g41240; CSD1, copper/zinc superoxide dismutase, At1g08830; putative prolyl oligopeptidase, At1g20380; lipid transfer protein related, At1g27950; Suc synthase, At4g02280; calcineurin-like phosphoesterase, At5g50400. Tubulin (At1g04820) was used as a control for equal cDNA use.

When comparing the other differentially expressed genes, TcAPX2, TcHMA4, and TcZIP4 (ZNT1) are all constitutively expressed in T. caerulescens leaves. Expression of these genes was either not detected in Arabidopsis leaves (APX2 and HMA4) or only detected at zinc deficiency conditions (ZIP4). Also, the expression of FER1 and FRO4 in leaves differs between the two species: TcFER1 is induced under zinc excess conditions while AtFER1 is induced under low zinc concentrations, and TcFRO4 is not expressed in the leaves while AtFRO4 is slightly induced at sufficient and excess zinc conditions. The HAK5 potassium transporter and the CSD1 copper/zinc superoxide dismutase genes show similar expression in leaves and roots of both species. In Arabidopsis, the At1g20380 gene (encoding a putative prolyl oligopeptidase) is strongly induced at low zinc conditions, whereas the T. caerulescens ortholog is more or less constitutively expressed at low levels in roots and leaves. ZIP1 is also constitutively lowly expressed in T. caerulescens roots and leaves, but induced in Arabidopsis roots at zinc deficiency. Furthermore, we found the IRT1 iron transporter gene to be very differently expressed in Arabidopsis compared to T. caerulescens. AtIRT1 is induced by excess zinc in Arabidopsis roots. For TcIRT1, this induction is much weaker in T. caerulescens roots, and the overall expression levels also are much lower than for the AtIRT1 gene, confirming the observed absence of this gene from Thlaspi cluster IV (Supplemental Table S7).

	DISCUSSION

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 
In this study, we investigated the expression of genes in roots of the nonaccumulator species Arabidopsis in response to exposure to three very different zinc concentrations in hydroponic culture. We postulate that genes that show differential expression under different zinc exposures are most likely to be involved in metal homeostasis. Most of these will be differentially expressed as a consequence of downstream changes in the physiological status of plants due to changes in metal homeostasis, but a few genes will be directly involved in regulating metal homeostasis. In trying to identify the latter ones, we also examined the differential expression of genes in the zinc hyperaccumulator T. caerulescens. Of the three metals we tested, only zinc homeostasis is clearly different between the two species. While Arabidopsis is not able to maintain nontoxic zinc levels in roots upon exposure to excess zinc levels in the nutrient solution (Fig. 1A; Becher et al., 2004; Talke et al., 2006), T. caerulescens is perfectly able to do this even while translocating high amounts of zinc to the leaves (Fig. 1A; Assunção et al., 2003b). After only 1 week, the zinc content in T. caerulescens is not as high as previously found by Assunção et al. (2003a), who measured after several weeks of exposure. Unexpectedly, iron accumulates in the roots of both T. caerulescens and Arabidopsis at increasing zinc concentrations (Fig. 1BI). Based on absence of an effect of iron status on zinc uptake in T. caerulescens (Lombi et al., 2002) and the antagonistic effect found for Arabidopsis seedlings (Thomine et al., 2003), we expected no effect or an antagonistic effect of the zinc status on iron uptake. The synergistic effect we found suggests that both species may increase their iron uptake as a response to a possible risk of iron deficiency in leaves. If so, this strategy is effective since no actual decrease is seen in iron concentration in T. caerulescens leaves and only a slight decrease is observed in Arabidopsis leaves (Fig. 1BII). For manganese, there is an antagonistic response of decreased uptake upon increased zinc uptake in the roots (Fig. 1C).

When examining gene expression in the same material of both species, we expected that genes that are differentially expressed between the two species, and especially those that show a difference in response to changes in the external zinc concentration, may be crucial to the adaptive difference between a zinc accumulator and a nonaccumulator. Previously, aspects of the metal accumulator versus nonaccumulator gene expression comparison have been studied for Arabidopsis and A. halleri (Becher et al., 2004; Weber et al., 2004; Talke et al., 2006). In addition to those studies, we compared and verified root gene expression under three different zinc exposure conditions (deficient, sufficient, and excess zinc) that clearly lead to different zinc concentrations in roots and leaves of both species. Another important addition to previous studies is that the Agilent 3 oligo microarray we used contains approximately 40,000 probes representing more than 27,000 annotated genes and more than 10,000 nonannotated transcripts (http://www.chem.agilent.com) and is thus an almost complete representation of the Arabidopsis transcriptome. We propose that genes that are induced in expression upon transfer to zinc deficiency or upon transfer to excess zinc are most interesting for further understanding of zinc homeostasis in Arabidopsis. Among the first class are some genes already known to be involved in zinc homeostasis, such as ZIP2, 4, 5, and 9, NAS2, and HMA2 genes (Grotz et al., 1998; Wintz et al., 2003; Talke et al., 2006). In addition, we confirmed previous suggestions of genes to be involved in zinc homeostasis, such as ZIP1, 3, and 10, IRT3, MTP2, and NAS4, to be more highly expressed under zinc deficiency. ZIP1, NAS2, and NAS4 were also induced in A. halleri in response to low zinc supply (Becher et al., 2004). Our results now suggest there are at least 10 different members of the ZIP gene family (Guerinot and Eide, 1999) that play a role in zinc uptake in roots (ZIP1, 2, 3, 4, 5, 9, 10, 11, and 12, and IRT3). Hypothesizing that these transporters are involved in transport of cations across the plasma membrane, it is unlikely that all of them are involved in the uptake of zinc in the same tissue. Most likely, these transporters exert a similar function in different parts of the root or are located in intracellular membranes.

Of the other two known zinc transporters induced by zinc deficiency, HMA2 has been implicated in transport of zinc into the vasculature, either to promote zinc export from root to shoot via the xylem or from shoot to root via the phloem (Eren and Argüello, 2004; Hussain et al., 2004). In the first case, higher expression under zinc deficiency could be a response to a higher zinc demand from the shoot; in the latter, it could be to accommodate the higher zinc demand of the root by remobilizing zinc from the shoot. The very strong induction of MTP2 is remarkable. Rather than MTP1 (previously known as ZAT), which is constitutively expressed in Arabidopsis (van der Zaal et al., 1999; Kobae et al., 2004), the induction of MTP2 by zinc deficiency suggests a specific role of this transporter in counteracting the effect of zinc deficiency.

Metals are often chelated in planta to NA. The absence of NA has severe effects on metal homeostasis, as was observed in the chloronerva mutant of tomato (Lycopersicon esculentum; Ling et al., 1999) or in NA metabolizing NAAT-overexpressing tobacco (Nicotiana tabacum) plants (Takahashi et al., 2003). NA is formed by trimerization of S-adenosylmethionine catalyzed by the enzyme NAS. Arabidopsis contains four NAS genes, at least three of which are able to catalyze the last step in the synthesis of NA (Suzuki et al., 1999; Becher et al., 2004; Weber et al., 2004). Only NAS2 and NAS4 are more highly expressed in roots under zinc deficiency compared to sufficiency (Fig. 3A), but the presence of several, apparently paralogous NAS genes with different overlapping gene expression profiles suggests complementary and possibly redundant functions.

In addition to the higher expression of NAS genes, some YSL genes also are induced by zinc deficiency. These genes are implicated in transport of metal-NA chelates within the plant (Curie et al., 2001; Waters et al., 2006) and possibly the entry of metals into the phloem or xylem (DiDonato et al., 2004). We find expression of YSL2 and YSL3 to be only slightly affected by different zinc treatments, and contrary to the observations by Schaaf et al. (2005), we find the genes to be slightly induced by lower zinc concentrations. Recently, Waters et al. (2006) also showed there is an induction of YSL3 in Arabidopsis grown under zinc deficiency conditions. Unexpected was the high zinc deficiency-induced expression of FRD3, FRO4, and FRO5. Although the frd3 mutant has a zinc accumulation phenotype, FRD3 has been mainly implicated in iron homeostasis (Lahner et al., 2003; Green and Rogers, 2004). FRO4 and FRO5 resemble the ferric chelate reductase gene FRO2 (Robinson et al., 1999), but, in contrast to FRO2, their expression is not induced in Arabidopsis roots upon iron deficiency (Mukherjee et al., 2005; Wu et al., 2005). The current results suggest a much broader role in general metal homeostasis of these genes than previously thought.

In addition to these genes, we identified 328 other probes with a similar differential transcription profile (Table II; Supplemental Table S3), indicating a similar involvement in zinc homeostasis for the corresponding genes (Eisen et al., 1998). For some of these probes, the only indication of a corresponding gene came from massive parallel sequence signature analysis (Meyers et al., 2004), as no Arabidopsis gene was annotated at that position. Three of these are among the 10 most differentially expressed when comparing zinc deficient and sufficient conditions. So far we have not identified the corresponding genes. For three other genes in this cluster, a prolyl oligopeptidase (At1g20380), a calcineurin-like phosphoesterase (At5g50400), and a bHLH family protein (At1g71200), knockout (KO) mutants were examined but not found to display any aberrant phenotype under differential zinc exposure (data not shown). Prolyl oligopeptidase and calcineurin-like phosphoesterase need metals, possibly also zinc, to function properly. The same holds for the iron-sulfur cluster assembly protein and carbonic anhydrase 1. This can explain their zinc-responsive expression profile.

Another large cluster of 128 differentially expressed genes is more highly expressed upon exposure to zinc excess. Expression of many of these appears to be associated with the defense against oxidative stress caused by this treatment (e.g. peroxidase, respiratory burst oxidase proteins). This cluster also comprises genes of families that are associated with iron deficiency response, such as ZIP genes (IRT1, IRT2, ZIP8), FRO genes (FRO1, 2, 3), MTPs (MTP3, 8), a NAS gene (NAS1), an oligopeptide transporter (OPT3), and IREG2. A large fraction of these was also found to be differentially expressed in the comparison between wild-type Arabidopsis and the fit1 mutant (Colangelo and Guerinot, 2004). The fit1 mutant is defective in a bHLH transcription factor controlling several genes involved in iron deficiency response. When comparing the list of 72 genes of which the expression is (partially) dependent on FIT1 (Colangelo and Guerinot, 2004) with the zinc excess-induced cluster (Table I; Supplemental Table S2), there are 30 genes in common. This apparent interaction between zinc and iron homeostasis in Arabidopsis, with zinc excess leading to iron deficiency, is not supported by a clear decrease in iron concentration in Arabidopsis leaves (Fig. 1), suggesting that this change in gene expression is indeed effective in avoiding actual iron deficiency in leaves.

For T. caerulescens, the zinc deficiency and zinc excess response is slightly different from Arabidopsis. This does not seem to be due to technical hybridization differences. The expression of the T. caerulescens genes confirmed by RT-PCR corresponded very well with the results obtained from the microarray analysis (Fig. 3; Supplemental Table S13). One cluster of coregulated genes is clearly differently expressed in T. caerulescens compared to Arabidopsis (Table III; Supplemental Table S5). Whereas in Arabidopsis the three ZIP family members ZIP3, ZIP4, and ZIP9 are only more highly expressed under zinc deficiency, their T. caerulescens orthologs are also relatively highly expressed under sufficient zinc conditions. Based on sequence similarity, the T. caerulescens ZNT1 and ZNT2 genes appear to be the orthologs of the Arabidopsis ZIP4 and IRT3 genes. Both were previously found to be very highly expressed in T. caerulescens, almost regardless of the zinc concentration in the medium (Pence et al., 2000; Assunção et al., 2001). Also, under zinc deficient conditions, these two T. caerulescens genes are much more highly expressed than their Arabidopsis orthologs (Table IV; Supplemental Table S12).

Comparable to Arabidopsis, T. caerulescens expresses a cluster of genes in response to zinc deficient conditions (Table III; Supplemental Table S5), although this cluster is much smaller than in Arabidopsis. Such might be caused by differences in hybridization efficiency, but this is probably not the case, as the T. caerulescens orthologs of FRD3, ZIP10, and HMA4 are not significantly differentially expressed within T. caerulescens, even though they are much more highly expressed in T. caerulescens than in Arabidopsis (Table IV; Supplemental Table S12). In a recent microarray study, FRD3 and HMA4 also appeared to be constitutively more highly expressed in A. halleri compared to Arabidopsis (Talke et al., 2006). Similarly, the expression of the NAS2 and FER1 orthologs in T. caerulescens is more or less constitutive rather than zinc deficiency induced as in Arabidopsis (Fig. 3).

The strong expression of NAS2 in A. halleri compared to Arabidopsis (Weber et al., 2004; Talke et al., 2006) was not found in T. caerulescens. The different expression profiles between Arabidopsis and T. caerulescens of the other three NAS genes (Fig. 3) suggest a major function for these genes in the metal adaptation of T. caerulescens. The presence of at least four NAS gene copies in both species, which are apparently all functional and highly redundant (no visible phenotypes are observed in soil-grown single and double KO Arabidopsis mutants; data not shown), will have provided ample flexibility to sustain adaptive changes in NAS gene expression. Also, in view of the observed effect on, especially, nickel tolerance upon NAS overexpression in nonaccumulating, nontolerant species (Douchkov et al., 2005; Kim et al., 2005; Pianelli et al., 2005), the NAS genes may be crucial to metal tolerance in T. caerulescens.

Most interesting for the identification of genes that contribute to the adaptation of T. caerulescens to high zinc exposure are the genes that are differentially expressed when comparing T. caerulescens and Arabidopsis at comparable zinc exposures (Table IV; Supplemental Table S10). More than 2,200 genes are significantly (P < 0.05) differentially expressed (5-fold) at any of the three zinc exposure treatments. This compares well with the recent transcript profile comparison of T. caerulescens and Thlaspi arvense shoot tissue, in which close to 3,500 genes were found to be more than 2-fold differentially expressed at P < 0.05 (Hammond et al., 2006). More than 50% of the genes we find more highly expressed in T. caerulescens are of unknown function. In a recent T. caerulescens EST analysis (Rigola et al., 2006), especially genes of unknown function were overrepresented while genes involved in general and protein metabolism were underrepresented when compared to Arabidopsis. Even though the fraction of genes involved in (a) biotic stress response is comparable in both species, the stress response genes expressed in T. caerulescens are generally different from those expressed in Arabidopsis.

We further analyzed this large set of species-specific differentially expressed genes in different ways. When sorting them according to the highest differential expression under zinc deficiency, which we consider to be most informative, there are several genes that are more than 100-fold higher expressed in T. caerulescens. Among the 15 most differentially expressed genes are four PDF or defensin genes, of which PDF1.1 is close to 1,000-fold higher expressed under zinc deficient and excess conditions. The biological role of defensins is not very clear. These small Cys-rich peptides are generally induced by fungal infections and implicated in pathogen defense, hence their name (Thomma et al., 2002). Mirouze et al. (2006) recently showed that the A. halleri PDF family confers zinc tolerance, and they hypothesized that defensins interfere with divalent metal cation trafficking to confer the zinc tolerance phenotype. In Arabidopsis, expression of PDF1.2 is induced by the stress hormone jasmonic acid (JA; Penninckx et al., 1998). Maksymiec et al. (2005) recently showed that heavy metal stress also induces JA accumulation in plants. Armengaud et al. (2004) found that PDF1.2a, PDF1.2b, PDF1.2c, and PDF1.3 are also among the most induced genes upon potassium starvation in Arabidopsis, and they suggested a relation between potassium starvation and JA signaling. Of the total 415 genes they found to be differentially expressed under potassium starvation, we found 46 genes to be more highly expressed in T. caerulescens compared to Arabidopsis. How or if JA, potassium starvation, and zinc response are correlated remains elusive. However, in this context it is also interesting to note that HAK5, KUP3, and KAT, three potassium transporter genes, are much more highly expressed in T. caerulescens roots compared to Arabidopsis.

Zinc hyperaccumulation is a constitutive trait in T. caerulescens and, thus, we expect it requires a constitutive expression of metal hyperaccumulation genes and no specific induction at zinc deficiency or excess. It is complicated to identify such zinc accumulation genes from the large set of more or less constitutively higher expressed T. caerulescens genes, as many genes in this large set will be involved in general species differences. However, when considering the 16 most highly expressed genes at 100 µM ZnSO₄, already six metal homeostasis genes are among them, four of which are known zinc transporters: HMA4, MTP1, and the already discussed ZIP4 and IRT3. HMA4 was previously identified by Papoyan and Kochian (2004) as a zinc-transporting, P-type ATPase possibly involved in zinc hyperaccumulation, particularly in loading of zinc into the xylem. The T. caerulescens ortholog of the Arabidopsis MTP1 gene was previously described as the ZTP1 gene (Assunção et al., 2001) and has been suggested to play a role in metal tolerance of T. caerulescens (Assunção et al., 2001) and Thlaspi goesingense (Persans et al., 2001; Kim et al., 2004). Paralogs of the AtMTP1 gene in A. halleri also cosegregate with zinc tolerance in a segregating population (Dräger et al., 2004). Other zinc transporter genes that are more highly expressed in T. caerulescens are HMA3, MTP8, and NRAMP3. HMA3 is a P-type ATPase similar to HMA4. When expressed in yeast, it is able to transport cadmium, but zinc transport could not be proven and in Arabidopsis the expression of the gene is not affected by exposure to zinc (Gravot et al., 2004). The MTP8 gene is another member of the cation diffusion facilitator family. Especially at zinc deficient and sufficient conditions, the gene is more highly expressed in T. caerulescens compared to Arabidopsis, suggesting a function in zinc uptake, although the regulation in Arabidopsis by FIT1 (Colangelo and Guerinot, 2004) also indicates a role in iron homeostasis. An mtp8 KO mutant was examined but not found to display any aberrant phenotype under differential zinc exposure (data not shown). AtNRAMP3 is a vacuolar transporter that is able to transport iron and cadmium but not zinc (Thomine et al., 2000). The specific induction of TcNRAMP3 gene expression by zinc deficiency and excess zinc suggests it plays an important role in mobilization of zinc and iron in T. caerulescens as in Arabidopsis (Lanquar et al., 2005).

Unexpected was that, in contrast to Arabidopsis, expression of the iron homeostasis genes IRT1, IRT2, and FRO2 is not induced in T. caerulescens upon excess zinc treatment. When we tested with RT-PCR, we could not detect the expression of TcIRT1 except in roots at lower zinc exposure levels (Fig. 3B; Supplemental Table S13). This suggests that T. caerulescens is either able to regulate zinc and iron homeostasis independently, unlike Arabidopsis, or that the continued expression of zinc transporters at high zinc exposure levels ensures low-efficiency, but sufficient, iron uptake in T. caerulescens. The IRT1 gene plays an interesting role in metal homeostasis in T. caerulescens, since it is very highly expressed upon iron deficiency in the cadmium-hyperaccumulating accession Ganges but much lower in the cadmium-excluding accession Prayon (Lombi et al., 2002). The latter is physiologically and geographically very close to La Calamine.

Two metallothionein genes, MT2B and especially MT2A, are very highly expressed in T. caerulescens. MT2 expression is generally associated with copper stress tolerance (Zhou and Goldsbrough, 1995; van Hoof et al., 2001), but overexpression in Vicea faba also induced cadmium tolerance (Lee et al., 2004). Why these two, more copper-associated genes are so highly expressed in T. caerulescens is unclear. High zinc uptake due to high expression of zinc transporters with a low affinity for copper could of course cause some copper stress, but high copper levels have not been reported for T. caerulescens. It is more likely that these genes have a function in general stress response or, alternatively, serve to maintain copper homeostasis, as was also suggested for the MT3 gene of T. caerulescens (Roosens et al., 2004).

Very surprising was the relatively high expression of 24 genes with a suggested function in lignin biosynthesis (Ehlting et al., 2005) and 13 genes implicated in suberin biosynthesis (CER3, CER6, and 11 LTP genes; Costaglioli et al., 2005) in T. caerulescens (Table IV). CER3 is known to be expressed in Arabidopsis roots, but the expression of CER6 in T. caerulescens roots is very different from the expression in Arabidopsis (Hannoufa et al., 1996; Hooker et al., 2002). The high expression of lignin/suberin biosynthesis genes coincides well with the progressed U-shaped lignification/suberinization of the endodermis cells and the occasional presence of a second endodermis layer found in T. caerulescens roots, but not in Arabidopsis roots (Fig. 2). Casparian strip development and lignification in cortical cells also was recently observed by Zelko et al. (2005) in T. caerulescens but not in the closely related nonaccumulator T. arvense. Comparable endodermis wall thickenings were also observed in the salt-adapted crucifer Thelungiella halophila (Inan et al., 2004). Strong deposition of lignin and suberin on the radial and inner tangential walls resulting in a U-like appearance of the endodermal cells is not uncommon for plants (Zeier and Schreiber, 1998). Since this cell wall deposition occurs most prominently at older parts of the root where root hairs are no longer active, we hypothesize that this layer acts to prevent excess efflux of metals from the vascular cylinder rather than to prevent uncontrolled influx.

With so many genes differentially expressed, one also expects alterations in the transcript levels of transcription factors. In the T. caerulescens-Arabidopsis comparison, we found 131 transcriptional regulators with more than 5-fold higher expression (FRD P < 0.05) in T. caerulescens (Supplemental Table S14). Of 19 genes that are more than 10-fold higher expressed under zinc sufficient conditions, two genes (INO and SPL) are associated with flower development in Arabidopsis (Villanueva et al., 1999; Yang et al., 1999) and expression is very atypical. However, in line with this atypical expression, we also found the FIS2 gene more highly expressed in T. caerulescens roots. In Arabidopsis, this gene is predominantly expressed in developing seeds (Luo et al., 2000), but also in A. halleri and Arabidopsis lyrata this gene is induced in roots in response to zinc exposure (www.genevestigator.ethz.ch).

In conclusion, the comparative transcriptional analysis of the hyperaccumulator T. caerulescens and the nonaccumulator Arabidopsis emphasizes the role of previously implicated zinc homeostasis genes in adaptation to high zinc exposure, but also suggests a similar role for many more, as yet uncharacterized genes, often without any known function. While some of these genes were also differentially expressed when comparing A. halleri with Arabidopsis, many more were not or at very different levels, suggesting that there will be overlap in the mechanisms of metal accumulation and metal tolerance but also many differences between metal hyperaccumulator species.

	MATERIALS AND METHODS

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 

Plant Material and Growth Conditions

Arabidopsis thaliana Columbia-0 (Arabidopsis) and Thlaspi caerulescens J. & C. Presl accession La Calamine seeds were germinated on garden peat soil (Jongkind BV). Three-week-old seedlings were transferred to 600-mL polyethylene pots (three plants per pot) containing a modified half-strength Hoagland nutrient solution (Schat et al., 1996): 3 mM KNO₃, 2 mM Ca(NO₃)₂, 1 mM NH₄H₂PO₄, 0.5 mM MgSO₄, 1 µM KCl, 25 µM H₃BO₃, 2 µM ZnSO₄, 2 µM MnSO₄, 0.1 µM CuSO₄, 0.1 µM (NH₄)₆Mo₇O₂₄, and 20 µM Fe(Na)EDTA. The pH buffer MES was added at 2 mM concentration and the pH was set at 5.5 using KOH. Each polyethylene pot contained three seedlings of T. caerulescens or Arabidopsis. Three weeks after growing on this solution, the T. caerulescens plants were transferred for 7 d to the same modified half-strength Hoagland nutrient solution with a deficient (0 µM), sufficient (100 µM), or excess (1,000 µM) ZnSO₄ concentration. The Arabidopsis plants were transferred to the same nutrient solution with deficient (0 µM), sufficient (2 µM), or excess (25 µM) ZnSO₄. During the first 3 weeks, the nutrient solution was replaced once a week and thereafter twice a week. Germination and plant culture were performed in a climate chamber (20°C/15°C day/night temperatures; 250 µmol light m^–2 s^–1 at plant level during 14 h/d [T. caerulescens] or 12 h/d [Arabidopsis]; 75% relative humidity; Assunção et al., 2001).

Root and Shoot Metal Accumulation Assay

Two pools of three plants, grown as described before, were used per treatment. After 4 weeks of growth, the plants were harvested, after desorbing the root system with ice-cold 5 mM PbNO₃ for 30 min. Roots and shoots were dried overnight at 65°C, wet-ashed in a 4:1 mixture of HNO₃ (65%) and HCl (37%) in Teflon bombs at 140°C for 7 h, and analyzed for zinc, iron, and manganese using flame atomic absorption spectrometry (Perkin-Elmer 1100B). Metal concentrations in roots and shoots were calculated as µmol per g DW.

Microarray Experiment

A common reference model was used to design the microarray experiment (Yang and Speed, 2002), in which cDNA from T. caerulescens roots exposed to 100 µM (sufficient) zinc was used as the common reference. Every slide was always hybridized with the common reference sample and with a sample from one of the treatments (Arabidopsis or T. caerulescens exposed to deficient, sufficient, or excess zinc). The common reference was labeled with the fluorescent dye Cyanine 3 and the treatment samples were labeled with Cyanine 5. As a quality-control step, we performed a dye-swap hybridization for one sample (from T. caerulescens roots exposed to sufficient zinc).

Roots of one pot containing three Arabidopsis or three T. caerulescens plants per treatment were pooled and homogenized in liquid nitrogen. Each pool was considered as one biological replicate and two biological replicates were used. Total RNA was extracted with Trizol (Invitrogen). Approximately 300 mg tissue was used for the RNA extraction performed according to the manufacturer's instructions. After the RNA extraction, total RNA was purified with RNeasy spin columns (Qiagen Benelux B.V.) and genomic DNA was digested with the RNase-free DNase set (Qiagen Benelux B.V.). Ten micrograms of total RNA was used to synthesize cDNA with MMLV reverse transcriptase and DNA primer. The cDNA was labeled with Cyanine 3-dCTP or Cyanine 5-dCTP and hybridized to Agilent Arabidopsis 3 60-mer oligonucleotide microarrays (Agilent Technologies) representing approximately 40,000 putative genes. The microarrays contain all Arabidopsis genes of known function and genes with a high degree of similarity to genes of known function from heterologous organisms. This includes more than 27,000 annotated genes and more than 10,000 nonannotated transcripts based on massive parallel sequence signature data (http://www.chem.agilent.com) and is thus an almost complete representation of the Arabidopsis transcriptome.

After hybridization, the slides were scanned, analyzed, and normalized with the Agilent Feature Extraction software (Agilent Technologies). The arrays were first normalized using Agilent's standard normalization within each array. The remaining statistical analysis was done using the limma package (Smyth, 2005a) in R/BioConductor (Gentleman et al., 2004). Between-array quantile normalization was performed on the common reference channel while leaving the log ratios unchanged (Yang and Thorne, 2003). To find differentially expressed genes, we performed a separate channel analysis (Smyth, 2005b) between the pairs of interest using a moderated t test (Smyth, 2004). This test is similar to a standard t test for each probe except that the SEs are moderated across genes to ensure more stable inference for each gene. This prevents a gene from being judged as differentially expressed with a very small fold change merely because of an accidentally small residual SD. The resulting P values were corrected for multiple testing using the Benjamini-Hochberg FDR adjustment (Benjamini and Hochberg, 1995). Genes were considered to be significantly differentially expressed if both the FDR P values were <0.05 (controlling the expected FDR to no more than 5%) and the fold change was 3 (within species) or 5 (between species). Genes found to be significantly differentially expressed were clustered using Cluster/Treeview (Eisen et al., 1998). Average linkage hierarchical clustering with uncentered correlation was used within Cluster to perform the clustering analysis. Primary microarray data are available in ArrayExpress E-MEXP-877.

Genomic DNA hybridizations were performed using 1 µg random primed genomic DNA. As a quality-control step, we performed a dye-swap hybridization. After hybridization, the slides were scanned, analyzed, and normalized with the Agilent Feature Extraction software (Agilent Technologies). The arrays were further normalized using linear Lowess analysis. The features that hybridized with Arabidopsis genomic DNA (both polarities of the dye swap) and did not hybridize with T. caerulescens genomic DNA were extracted from the dataset by Spotfire using a fold change 3 as the cutoff value.

Semiquantitative RT-PCR

Total RNA of leaves and roots of a third Arabidopsis or T. caerulescens biological replica was extracted with Trizol (Invitrogen). Selected T. caerulescens genomic and cDNA fragments were PCR amplified using primers designed for the orthologous Arabidopsis gene and cloned into the pGEM-T Easy vector (Promega). Clones were sequenced and new primers were designed for semiquantitative RT-PCR to ensure amplification of the correct T. caerulescens gene. Five micrograms of total RNA was used to synthesize cDNA with MMLV reverse transcriptase (Invitrogen) and oligo(dT) as a primer (Invitrogen). The PCR amplification was performed with a cDNA aliquot (1 µL) and gene-specific primers (Supplemental Table S13). Care was taken to design Arabidopsis primers at comparable positions and with comparable length and T_m as for T. caerulescens primers to allow proper comparison of the expression data. Primers for Tubulin (Supplemental Table S13) were used as a control for similar cDNA quantity between the samples. Between 25 to 35 PCR cycles (30 s at 94°C, 30 s at 50°C, and 60 s at 68°C) were performed in a 50-µL volume, preferably with the same number of cycles for Arabidopsis and T. caerulescens samples. Twenty microliters of the reaction was separated on an ethidium bromide-stained 1% agarose gel. Gel-image analysis using QuantityOne software (Bio-Rad) was used to quantify the DNA fragment intensities (Supplemental Table S13). The DNA fragment intensities were corrected for background signal and corrected for cDNA quantity using the intensities of Tubulin.

Microscopic Analysis of Lignification in T. caerulescens

T. caerulescens La Calamine seeds were germinated and plants were grown as described before on sufficient zinc medium. After 4, 6, and 9 weeks, roots of two plants were collected and hand sections were made by repeatedly chopping roots on a microscope slide using a razorblade. These sections were analyzed with a Nikon Labophot bright-field microscope.

T. caerulescen