ECA3, a Golgi-localized P2A-type ATPase, plays a crucial role in manganese nutrition in Arabidopsis.

Calcium (Ca) and manganese (Mn) are essential nutrients required for normal plant growth and development, and transport processes play a key role in regulating their cellular levels. Arabidopsis (Arabidopsis thaliana) contains four P2A-type ATPase genes, AtECA1 to AtECA4, which are expressed in all major organs of Arabidopsis. To elucidate the physiological role of AtECA2 and AtECA3 in Arabidopsis, several independent T-DNA insertion mutant alleles were isolated. When grown on medium lacking Mn, eca3 mutants, but not eca2 mutants, displayed a striking difference from wild-type plants. After approximately 8 to 9 d on this medium, eca3 mutants became chlorotic, and root and shoot growth were strongly inhibited compared to wild-type plants. These severe deficiency symptoms were suppressed by low levels of Mn, indicating a crucial role for ECA3 in Mn nutrition in Arabidopsis. eca3 mutants were also more sensitive than wild-type plants and eca2 mutants on medium lacking Ca; however, the differences were not so striking because in this case all plants were severely affected. ECA3 partially restored the growth defect on high Mn of the yeast (Saccharomyces cerevisiae) pmr1 mutant, which is defective in a Golgi Ca/Mn pump (PMR1), and the yeast K616 mutant (Δpmc1 Δpmr1 Δcnb1), defective in Golgi and vacuolar Ca/Mn pumps. ECA3 also rescued the growth defect of K616 on low Ca. Promoter:β-glucuronidase studies show that ECA3 is expressed in a range of tissues and cells, including primary root tips, root vascular tissue, hydathodes, and guard cells. When transiently expressed in Nicotiana tabacum, an ECA3-yellow fluorescent protein fusion protein showed overlapping expression with the Golgi protein GONST1. We propose that ECA3 is important for Mn and Ca homeostasis, possibly functioning in the transport of these ions into the Golgi. ECA3 is the first P-type ATPase to be identified in plants that is required under Mn-deficient conditions.

Calcium (Ca) and manganese (Mn) are essential nutrients required for normal plant growth and development. Both Mn deficiency and excess Mn accumulation can result in severely decreased crop yield. Mn deficiency is one of the most widespread trace element deficiencies for arable crops, often seen in cereals such as wheat (Triticum aestivum) and barley (Hordeum vulgare;Jiang, 2006). Plants deficient in Mn appear chlorotic and growth is reduced (Shenker et al., 2004). Mn 21 is essential throughout all stages of plant development; it activates a number of different enzymes, such as decarboxylases and dehydrogenases in the Krebs cycle (Marschner, 1995), and several glycosyltransferases in the Golgi (White et al., 1993;Nunan and Scheller, 2003). Mn 21 is associated with PSII, where it is required for light-induced water oxidation through which oxygen is produced. Mn is also required for mitochondrial superoxide dismutase (MnSOD), which is involved in scavenging of superoxides (Bowler et al., 1991) and is important in the biosynthesis of fatty acids and carotenoids (Marschner, 1995). As well as being an essential element, Mn 21 is also toxic at higher concentrations causing chlorosis, brown speckles on mature leaves, and necrosis (Marschner, 1995).
Transport proteins play a crucial role in maintaining the correct concentrations of Mn in the cytoplasm and also in supplying Mn to particular subcellular compartments where it may be required to act as an enzyme cofactor. In yeast (Saccharomyces cerevisiae) and also in many bacterial systems, several pathways for Mn 21 transport have been identified (Kehres and Maguire, 2003;Culotta et al., 2005). In plants, the situation is less clear, but several proteins that may transport Mn have been identified in Arabidopsis (Arabidopsis thaliana), including AtECA1, an endoplasmic reticulum 1 This work was supported by the Biotechnology and Biological Sciences Research Council (51/P17217) and European Union Framework VI program for Research and Technological Development (contract no. FOOD-CT-2006-016253). This publication reflects only the authors' views; the Community is not liable for any use that may be made of the information contained therein. * Corresponding author; e-mail l.e.williams@soton.ac.uk. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Lorraine E. Williams (l.e.williams@soton.ac.uk). [C] Some figures in this article are displayed in color online but in black and white in the print edition. [W] The online version of this article contains Web-only data. [OA] Open Access articles can be viewed online without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.107.110817 (ER) P 2A -type ATPase; AtCAX2, a vacuolar cation/H 1 antiporter; AtIRT1, a plasma membrane ZIP transporter; and AtNramp3, a vacuolar Nramp (for review, see Williams et al., 2000;Hall and Williams, 2003;Pittman, 2005). Certain members of the cation diffusion facilitator family also appear to transport Mn, including AtMTP11 (Delhaize et al., 2007;Peiter et al., 2007). Ca is required at much higher concentrations than Mn and plays many fundamental roles in plants, including stabilization of cell walls and membranes, facilitating root extension, and protection against potentially damaging factors in the soil, such as salinity (Sanders et al., 1999). Ca deficiency is rare in nature, but disorders that are observed in horticulture tend to be seen in fast-growing tissues and include blossomend rot in tomato (Solanum lycopersicum), tip-burn in leafy vegetables, and fruit cracking (Marschner, 1995). Ca is also important in signaling, coupling a wide range of extracellular stimuli to a vast array of intracellular responses (Sanders et al., 1999). As such, the cytosolic Ca concentration must be carefully regulated and resting levels are usually around 0.1-0.2 mM. Ca 21 enters the cytoplasm through calcium channels present at both the plasma membrane and intracellular membranes (White, 2002). Ca 21 efflux is mediated by high-affinity P-type Ca-ATPases and lower affinity Ca/H antiporters (Evans and Williams, 1998;Geisler et al., 2000, Cheng et al., 2005. The complexity of Ca signaling and homeostasis is reflected in the diversity of Ca 21 transporters. Molecular analyses of Ca-ATPases in plants indicate that they divide into two types: P 2A -type Ca-ATPases (also known as ECAs), which show homology to the sarcoplasmic/ER Ca-ATPases (SERCAs), and P 2B -type Ca-ATPases (also known as ACAs), which are generally stimulated by calmodulin and show homology to the calmodulin-binding Ca-ATPases found at the plasma membranes of animal cells (PMCAs; Axelsen and Palmgren, 1998;Geisler et al., 2000). A subset of P 2A -type Ca-ATPases, the secretory pathway Ca-transport ATPases (SPCAs), is found in other organisms (PMR1 is an example in yeast), but these do not appear to exist in plants.
Ca pumps may have a variety of functions: loading of Ca into intracellular stores, such as the ER and vacuole, for them to act as sources of Ca for signaling; shaping and terminating a Ca signal; providing Ca for vesicle trafficking and fusion; and supplying Ca and possibly other divalent cations, such as Mn and zinc (Zn), to organelles to support their biochemical reactions (Sze et al., 2000). Genetic studies using mutants are now starting to reveal specific functions for individual Ca pumps (Wu et al., 2002;Schiott et al., 2004).
There are four P 2A -type ATPases in Arabidopsis (AtECA1-AtECA4). Of these, AtECA1 was the first to be studied (Liang et al., 1997) and was shown to function at the ER in the transport of Ca and Mn (Liang et al., 1997;Liang and Sze, 1998). The growth of the eca1-1 mutant is sensitive to low Ca and high Mn, suggesting that pumping of these cations into the ER is required to support plant growth under conditions of Ca deficiency and Mn toxicity (Wu et al., 2002). Having shown that additional P 2A -type ATPases, ECA2 and ECA3, are expressed in Arabidopsis (Pittman et al., 1999), the main objective of this study was to investigate their physiological role. This was achieved by Figure 1. Isolation of mutant lines. A, Determining the genotype of plants from Salk line 045567. PCR was carried out on genomic DNA from individual plants (1-3) using gene-specific primers ECA3Ex10.F and ECA3.R to detect plants that contained a wild-type copy of the gene; these amplified a band of the predicted size, 2.07 kb (top). Salk LBa1 primer for the T-DNA and the gene-specific primer ECA3.R were used to detect the presence of the T-DNA in ECA3. They amplified a band of the approximate predicted size, 1.96 kb (bottom). The zygosity of plants 1 to 3 determined from this analysis is indicated: wt, Wild-type; hom, homozygous for the insert; het, heterozygous for the insert. B, RT-PCR using RNA isolated from wild type and eca mutants. Primers designed to amplify a product across the insertion site gave products for ECA2 and ECA3 of 0.67 and 1.67 kb, respectively, only in wild-type plants. Actin 2 primers were used to amplify a product of 0.2 kb as a positive control in all samples. C, Insertion sites for mutants isolated in this study.
isolating T-DNA insertion mutants in which the genes encoding these proteins had been disrupted and comparing the phenotype of the mutants to wild-type Arabidopsis. We present evidence showing that eca3 mutants have a unique and distinct phenotype from eca1 and eca2 mutants and demonstrate that ECA3 has a crucial role in Mn nutrition.

Primary Structure of Arabidopsis ECA3
AtECA3 from ecotype 'Columbia-0' contains an open reading frame of 998 amino acids with a predicted molecular mass of 109.062 kD. It is identical to the sequence predicted from genome sequencing (The Arabidopsis Information Resource; http://arabdopsis. org), but differs from the ECA3 sequence (AJ132388) that we cloned previously from 'Landsberg erecta' (Pittman et al., 1999) in four bases: A336G, T453A, T1647G, and C2934T. In the ECA3 sequence from 'Landsberg erecta', the second-and fourth-base changes are silent, but the first and third give changes in the amino acid sequence: R46G and S549R. Conpred II, a consensus prediction method for obtaining transmembrane (TM) topology models (http://bioinfo.si. hirosaki-u.ac.jp/;ConPred2), indicates that AtECA3 contains 10 TM domains, a small cytoplasmic loop between TMs 2 and 3 and a large cytoplasmic loop between TMs 4 and 5. The latter contains the highly conserved phosphorylation domain CSDKTGTLT, which includes the phosphorylated Asp residue (Asp-347 for AtECA3) that is phosphorylated during the reaction cycle in all P-type ATPases. This model is consistent with structural models that have been presented for the sarcoplasmic reticulum Ca-ATPase from animal cells (Toyoshima et al., 2000).

Isolation of T-DNA Insertion Mutant Alleles of ECA2 and ECA3
To determine the physiological function of ECA2 and ECA3, we obtained mutant lines in which ECA2 and ECA3 were likely to be functionally disrupted. We isolated two independent mutant alleles for each (eca2-1, eca2-2, eca3-1, and eca3-2). These T-DNA insertion alleles were identified using the reverse-genetics approach outlined in ''Materials and Methods.'' Figure  1A illustrates genotyping of plants from the Salk line 045567; Figure 1B shows that, for each particular mutant, no product with the appropriate gene-specific primers could be detected in reverse-transcribed cDNA from plants homozygous for the insert. Figure  1C and Table I summarize the T-DNA insertion positions in the resulting mutant alleles and indicate where in the protein the insertion would occur. The position of the inserts in each of the mutants means that these are predicted loss-of-function mutants.

Effect of Low Ca on eca Mutants
To elucidate the biological function of ECA2 and ECA3 in Arabidopsis, we monitored insertion mutants under a wide range of conditions comparing their growth and development to wild-type plants. When grown on soil, no major differences from wild-type plants were observed (Supplemental Fig. S1).
Because eca1-1 mutants were previously shown to be extremely retarded when grown on low-Ca medium (0.2 mM; Wu et al., 2002), eca2 and eca3 mutants were tested in this range. At basal levels of Ca (1.5 mM), eca3-1, eca3-2, eca2-1, and eca2-2 mutants were similar in their fresh weight and chlorophyll content to wildtype plants; results are shown only for fresh weight for eca2-1 and eca3-1 ( Fig. 2A). In contrast to results reported previously for eca1-1 mutants (Wu et al., 2002), eca2 and eca3 mutants were no more retarded in growth than wild-type plants at low Ca (in this case, 0.1 mM). We next tested whether there was a difference in the growth of the mutants when Ca was omitted from the medium. Under these conditions, both wildtype and mutant seedlings were adversely affected ( Fig. 2B). They became chlorotic and stopped growing at an early stage. In general, both eca3 mutants were slightly more sensitive than wild-type and eca2 mutants, but in this experiment only eca3-1 was significantly lower in fresh weight (P , 0.05; Fig. 2B).
eca3 Mutants Are Susceptible to Growth in Mn-Deficient Conditions eca1-1 mutants were previously shown to be more adversely affected in growth on high (0.5 mM) Mn medium (Wu et al., 2002). We found no difference in the response of eca2 and eca3 mutants to this level of Mn compared to wild-type plants (data not shown). Therefore, we investigated whether the mutants showed any responses under Mn-deficient conditions. A striking difference was observed in the eca3 mutants compared to eca2 and wild type when grown on Table I. eca mutant alleles identified in this study Introns and exons are numbered so that the first exon begins with the start ATG and is followed by the first intron. medium with Mn omitted (Fig. 3). Under these conditions, eca3 mutants showed a marked reduction in growth, which first became apparent after approximately 8 to 9 d. Both root and shoot growth were reduced (Fig. 3, A and B). In addition, the cotyledons became yellow and the marked reduction in chlorophyll is indicated in Figure 3B. There were some interesting features observed when individual seedlings were examined closely ( Fig. 3A; Supplemental Fig. S2).
In some seedlings, dark areas were seen at the tips of developing leaves (Supplemental Fig. S2). On leaves where the lamina was expanding, a brown band was often observed across the petiole or lamina with bleaching occurring distal to the band. In many cases, the leaf lamina failed to expand further and senescing areas were observed. When leaves did develop, these and the expanded cotyledons were seen to bleach. Addition of Mn to the medium at a concentration of around 1 mM was sufficient to suppress these phenotypes (Fig. 4).
The response of Arabidopsis mutants to other ion deficiencies was also investigated. When Zn was omitted from the growth medium, there was no significant difference in the performance of the eca3 (Fig.  5A) or eca2 mutants (data not shown) compared to wild type. eca3 mutants were inhibited in their growth to the same extent as wild type and there were no significant differences in terms of fresh weight of shoot or chlorophyll content. Omission of iron (Fe) from the medium had a marked effect on growth and chlorophyll content of all plants, but there was no specific detrimental effect on the mutants compared to wildtype plants (Fig. 5B). Omitting copper (Cu) from the medium also had no specific effect on mutants compared to wild type (Fig. 5B).

Elemental Analysis of eca3 Mutants
Because the eca3 mutants showed a clear phenotype under Mn-deficient conditions, Mn content was determined under nutrient-sufficient (basal; 50 mM Mn) and Mn-deficient (0 mM Mn) conditions. The levels observed in each genotype are shown in Figure 6. To make this clearer, the level relative to the wild type (expressed as a percentage) is shown in Supplemental Figure S3, with the level in the wild type (100%) indicated as a dashed line. Despite the marked differences observed in the growth phenotype of eca3 mutants compared and eca3-1 (bottom) compared to wild-type (wt) in A and for wild type and all mutants in B; data represent the means 6 SE from five separate plates, each having six seedlings per genotype per plate (wild type, black bar; mutant, gray bar). *, Significant difference between mutant and wild-type plant (P , 0.05); Student's t test (paired). Results are from representative experiments repeated at least twice for each mutant. [See online article for color version of this figure.] to wild-type plants when grown under Mn-deficient conditions, there were no major differences in metal content changes when eca3 mutants and wild-type plants were compared. Generally, Mn content of eca3 mutants was slightly lower than wild-type plants under both conditions (i.e. basal Mn supplied and Mn omitted). Ca content was slightly higher under both conditions in the eca3 mutants compared to wild-type plants and Figure 3. eca3 mutants show reduced growth and chlorosis on Mn-deficient medium. Plants were grown on plates in 0.53 Murashige and Skoog* medium containing normal levels of Mn (50 mM) or without Mn present (0 mM) for 17 d. A, Seedling phenotype of eca2-1, eca3-1, and eca3-2 mutants compared to wild type (wt). Black line 5 1 cm. B, Fresh-weight and chlorophyll content for eca3-1 and eca3-2 mutants (gray bars) compared to wild type (black bars). C, Fresh-weight and chlorophyll content for eca2-1 mutant (gray bars) compared to wild-type (black bars). Results in B and C represent the means 6 SE from at least five separate plates, each having six seedlings per genotype per plate. *, Significant difference between mutant and wild-type plants (P , 0.05); Student's t test (paired). Results are from a representative experiment repeated at least three times. [See online article for color version of this figure.] Zn was similar or slightly lower (Fig. 6). Again, for both Ca and Zn, there was little difference in the response to reducing the Mn concentration in the medium of eca3 mutants compared to wild-type plants. We cannot rule out that extended growth under Mn-deficient conditions could lead to metal content changes, but we chose a point at which the symptoms, although apparent, were not too severe.

Expression Analysis of Type 2A Ca-ATPases
Semiquantitative reverse transcription (RT)-PCR was used to determine whether ECAs were present in various organs of hydroponically grown Arabidopsis. Primers designed to regions of AtECA1 to AtECA4 amplified products of 127, 325, 279, and 127 bp, respectively (Fig. 7A). Amplification of a 122-bp product with primers to the constitutively expressed ribosomal protein S16, a component of the 40S subunit (Kim et al., 2006), was used as a control. Transcripts for all ECAs are present in all major organs (Fig. 7A). Results for ECA2 and ECA3 are consistent with data from publicly available microarray studies (https:// www.genevestigator.ethz.ch; Zimmermann et al., 2004), which also show that these genes are expressed in all major organs (Fig. 7B). Expression of ECA1 and ECA4 cannot be determined from microarray analysis because the probe sets are not distinct.
To confirm these results and obtain further information about the tissue-specific expression of ECA3 within the different plant organs, a GUS reporter gene was cloned under the control of the promoter region for ECA3. GUS staining of seedlings showed high expression levels in guard cells, hydathodes, and the vascular tissue in leaves (Fig. 8). In primary roots, high levels of expression were seen at the root tip and vascular system. Transverse sections of roots showed high GUS activity in the stele, particularly in pericycle and xylem parenchyma cells. Staining was also observed in developing lateral roots and tip and vascular staining was seen in more mature lateral roots. Strong staining was also found in different flower tissues, including stamens, petals, and sepals, as well as in siliques (Fig. 8).
Some metal transporters are regulated at the transcriptional level by the metals they transport. For example, AtHMA4, a Zn-transporting ATPase, is upregulated by elevated levels of Zn (Mills et al., 2003;Williams and Mills, 2005), and the Fe transporter IRT1 is up-regulated under Fe-deficient conditions (Vert et al., 2001). Others, such as MTP1/ZAT, show no  . eca mutants show a similar response to wild-type plants when grown in the absence of Zn, Fe, or Cu. A, Seedling phenotype of eca3-1 and eca3-2 mutants compared to wild type (wt) when grown for 27 d on plates in 0.53 Murashige and Skoog* medium with Zn present at the normal concentration (15 mM; 1Zn) or absent (2Zn). Fresh weight of shoot (top) and chlorophyll content (bottom) for eca3-1 and eca3-2 mutants (gray bars) compared to wild type (black bars). B, Wildtype (black bars) and eca mutants (gray bars) were grown for 19 d on either 0.53 Murashige and Skoog medium (control) or in the absence of either Fe (2Fe) or Cu (2Cu). Data in A and B generally represent means 6 SE from five separate plates, each having six seedlings per genotype per plate. *, Significant difference between mutant and wildtype plant (P , 0.05); Student's t test (paired). obvious regulation by metals at the level of transcription (van der Zaal et al., 1999). Real-time PCR was performed to determine whether expression of ECA2 and ECA3 changed under conditions of metal deficiency. Root and shoot expression levels of ECA2 and ECA3 relative to the S16 component of ribosomal 40S are shown for seedlings grown on basal medium (control) or in the absence of either Mn or Zn (Fig. 9). There were no significant differences in the expression of either gene when grown under these conditions. For a comparison, we measured expression levels of the transporter ZIP4 and in this case there was marked upregulation under Zn-deficient conditions as reported previously (Wintz et al., 2003;van de Mortel et al., 2006).

ECA3 Is Localized to the Golgi
To determine the subcellular localization of ECA3, fusions with yellow fluorescent protein (YFP) were generated. Transient expression in tobacco (Nicotiana tabacum) epidermal cells followed by confocal laserscanning microscopy revealed small motile organelles characteristic of Golgi (Fig. 10). There was a high level of colocalization with the Golgi protein GONST1 (Handford et al., 2004) visualized using GONST1-GFP fusions.   analysis of plant Ca pumps (Harper et al,. 1998;Liang and Sze, 1998). Growth of K616 yeast is inhibited in the presence of elevated Mn or Zn in the medium (Wu et al., 2002). ECA3 rescues K616 on high Zn, restoring growth almost to the level of control K601 cells (Fig.  11A). ECA3 also partially restored growth of K616 on elevated Mn, but not to the extent of the K601 cells. Increased sensitivity to elevated Mn concentrations is also observed in the pmr1 mutant (Lapinskas et al., 1995) and ECA3 reduced this growth defect compared to vector controls (Fig. 11C). K616 vector transformants are also unable to grow on medium containing reduced levels of Ca (i.e. in the presence of EGTA). ECA3 suppresses the EGTA-sensitive phenotype, allowing growth in the presence of EGTA to levels similar to those seen in the control K601 strain (Fig. 11B).

DISCUSSION
Arabidopsis contains four P 2A -type ATPases (ECA1-ECA4), which are related to the well-characterized SERCA Ca-ATPases in mammals, and 10 P 2B -type ATPases related to the mammalian plasma membrane calmodulin-stimulated Ca 21 -ATPases. These are all members of the P-type superfamily of ATPases. Understanding the functional properties, expression patterns, and localization of each of these pumps is important if we are to understand how they contribute to ion homeostasis and signaling in plants. The aim of this study was to gain insight into the physiological roles of ECA2 and ECA3 by studying Arabidopsis T-DNA insertion mutants in which the genes encoding these ATPases have been disrupted. In addition, we have used heterologous expression in yeast to study functional properties of ECA3.
The studies conducted here with eca3 mutants have revealed a crucial role for ECA3 in Mn nutrition. eca3 mutants showed poor growth on Mn-deficient medium compare to wild-type plants ( Fig. 3; Supplemental Fig. S2). The response to Mn deficiency is first seen after about 8 to 9 d on this medium; prior to this, growth is similar to wild type. Omitting Zn, Fe, or Cu from the medium had little specific effect on the mutant compared to wild type (Fig. 5). In the absence of Mn, eca3 mutants showed a reduction in both root and shoot fresh weight and in chlorophyll content; this phenotype could be rescued by adding 1 mM Mn to the medium (Fig. 4). The fact that eca3 mutants show little difference from wild-type plants when Mn was present in the medium at normal levels indicates that other genes can compensate for the loss of ECA3 under these conditions. Notably, on soil there were no marked differences in the development of eca3 mutants, which grew similarly to wild type; in this case, there was probably sufficient Mn for normal growth and development. There appear to be distinct differences in the roles ECAs play in Mn and Ca tolerance. Wu et al. (2002) found that eca1 mutants were sensitive to high Mn and suggested that ECA1 may confer tolerance to this metal by removing it from the cytosol. Physical symptoms of Ca deficiency have also been reported in eca1-1 mutants when external Ca was reduced to 0.2 to 0.4 mM, suggesting that ECA1 may be the main pump capable of loading Ca into the ER lumen under these conditions (Wu et al., 2002). Both eca2 and eca3 mutants grew normally in this range, but, when Ca was omitted from the medium altogether, eca3 mutants seemed slightly more affected than eca2 mutants or wild type, suggesting a role for ECA3 in Ca homeostasis (Fig. 2). However, it should be stressed that, under these conditions, all plants were severely affected and the differences in the mutants and wild type were not as marked as seen for the situation when Mn was omitted from the medium.
Results from semiquantitative RT-PCR and microarray analysis showed that ECA3 is expressed in all major organs of Arabidopsis (Fig. 7). Promoter:GUS studies confirmed this and showed that ECA3 is highly expressed in a range of tissues and cells, including root tips, hydathodes, guard cells, and vascular tissue (Fig. 8).
The markedly different metal-dependent phenotypes observed in eca1 mutants (Wu et al., 2002) and eca3 mutants (this study) indicate that these ATPases have different physiological roles in Arabidopsis. This could be related to differences in their membrane location. ECA1 is associated primarily with the ER functioning in the transport of Ca and Mn into this compartment (Liang et al., 1997). Recent proteomic analysis (Dunkley et al., 2006) confirms ER localization for ECA1 and Golgi localization for ECA3 (P. Dupree, unpublished data). Using transient expression of YFP-ECA3 fusion proteins in tobacco, we show here that ECA3 colocalizes with GONST1 (Fig. 10), a protein previously shown to be located in the Golgi (Handford et al., 2004).
From the data we have presented in this article, we propose that ECA3 could serve a similar function to Golgi-associated SPCAs found in other organisms. In mammalian cells, there are three distinct classes of P 2 -type Ca-ATPases, SERCA type, PMCA type, and SPCAs (also referred to as PMR1 type), and these three groups can be clearly distinguished in phylogenetic analysis (Pittman et al., 1999;Schmidt et al., 2002;Supplemental Fig. S4). SPCAs transport Ca, but also Mn, with high affinity, and they function both in Mn detoxification via exocytosis and also in providing Mn for Golgi-localized enzymes (Yadav et al., 2007). In yeast, the SPCA, PMR1, functions as a Ca and Mn transporter at the Golgi, transporting these ions into the secretory pathway. Both ions can exert distinct functions: Ca supplied by PMR1 is required to sustain vacuolar protein sorting, whereas Mn is required to activate various Mn-requiring enzymes involved in the addition of complex carbohydrates onto N-and O-linked glycosylated proteins (Lapinskas et al., 1995;Durr et al., 1998). The fact that, on low Ca medium, ECA3 complements the yeast triple null mutant strain K616 (Dpmr1 Dpmc1 Dcnb1), lacking its endogenous Golgi (PMR1) and vacuolar (PMC1) Ca pumps, is consistent with ECA3 functioning as a Ca pump in yeast (Fig. 11). K616 also exhibits growth sensitivity to high levels of extracellular Mn, which results from toxicity of excessive cellular levels due to the lack of functional endogenous pumps (Lapinskas et al., 1995, Culotta et al., 2005. Expression of ECA3 reduces the Mn-sensitive phenotype of the triple yeast mutant K616 and also the single pmr1 mutant, indicating that ECA3 could serve a similar function to PMR1 (Fig. 11). ECA3 is particularly effective in reducing the Znsensitive phenotype in yeast, suggesting that it may  have the capability of transporting Zn. However, ECA3 does not seem to play a major role in Arabidopsis under Zn deficiency because wild-type plants and eca3 mutants were similarly affected under such conditions.
There are SPCAs in fungi, animal, and bacteria, but they have not been identified in higher plants. Phylogenetically, AtECA3 clusters with animal-type SERCA pumps, whereas AtECA1 and AtECA2 cluster more closely with plant-type SERCA pumps (Pittman et al., 1999;Supplemental Fig. S4). Our data indicate that, in plants, ECA3-type pumps could have a comparable physiological role to SPCA types and function in transporting Ca and Mn into the Golgi-associated secretory pathway. The partial rescue of metal-dependent growth phenotypes displayed in yeast mutants would support a transport function for ECA3, but more direct transport assays would be required to show that ECA3 functions as a Ca and Mn transporter. Future experiments will be aimed at testing the hypothesis that ECA3 transports Mn into the Golgi to supply various Mn-requiring enzymes for biochemical processes required for normal growth of Arabidopsis.
Mn deficiency is a serious nutritional disorder for plants in many areas of the world, and in arable crops this can lead to reduced yield. Not only is there variability between plant species in their ability to grow on low Mn, but there are also considerable differences among genotypes of the same species (Pedas et al., 2005). Varieties displaying tolerance to Mn deficiency have been termed Mn efficient (Pedas et al., 2005;Jiang, 2006), but the mechanisms underlying this have not been clearly defined. In barley, differential capacity for high-affinity Mn influx is thought to contribute to Mn efficiency; in wheat, it may be due to improved internal utilization rather than higher Mn accumulation (Jiang, 2006). ECA3 homologs could play a role in the Mn efficiency of crop species and there is the potential in the future to test whether expression of AtECA3 could improve Mn efficiency.
In summary, this study has provided evidence that AtECA3 plays a crucial role in Mn nutrition in Arabidopsis. Golgi localization of ECA3 could explain why disruption of ECA3 does not lead to a major effect on the overall Mn content of the plant. The severe phenotype observed under Mn-deficient conditions may be due to a reduction in the Mn content of the Golgi. Our observations indicate that other secretory pathway/endomembrane-localized Mn transporters, such as AtMTP11, cannot compensate for the lack of ECA3 under Mn deficiency. AtECA1, AtCAX2, and AtMTP11 have previously been reported to confer tolerance to high-Mn concentrations, but AtECA3 is the first to be identified that is required for growth under low-Mn conditions.
(Melford), 1% (w/v) Suc, and either one-half-strength (0.53) Murashige and Skoog medium (Murashige and Skoog, 1962) or the same medium but with Ca levels reduced from 1.5 mM to 0.1 mM (0.53 Murashige and Skoog*). For experiments investigating the effect of Mn deficiency on phenotype, Mn was omitted from the 0.53 Murashige and Skoog* medium. Seeds were stratified at 4°C for 48 h prior to transfer to a controlled-environment cabinet (23°C, 16 h light; 18°C, 8 h dark) and plates were incubated vertically or horizontally as indicated. To examine the development of the plants to maturity, seeds were sown in soil (one part Levingtons F5 no. 2; one part John Innes no. 2; one part Vermiperl-graded horticultural vermiculite, medium grade) sterilized by autoclaving at 121°C for 15 min at 1 bar pressure, and plants were grown in a controlled-environment cabinet (22°C, 16 h light; 20°C, 8 h dark cycle) or under glasshouse conditions. To obtain organ material from mature plants for gene expression analysis, plants were grown hydroponically as described previously (Mills et al., 2003).

Chlorophyll Determination and Fresh-Weight Measurements of Seedlings
Fresh-weight and chlorophyll measurements were determined using seedlings grown on five separate plates, each plate having six wild-type seedlings and six mutant seedlings. The data presented are generally the means from the five plates 6SE expressed on a per-seedling basis. Student's t test (paired) was used to determine significant differences between mutant and wild-type plants and an asterisk (*) indicates a significant difference (P , 0.05). Chlorophyll was determined following extraction in N,N-dimethylformamide (Moran, 1982).

Elemental Analysis
Shoots and roots were harvested separately from plants grown on agarose plates (see above). Plant samples were dried for 2 d at 70°C and weighed. Samples were digested at 120°C in 0.5 mL of 65% nitric acid for 4 h. After the solution was cooled, it was diluted to 10 mL with deionized water and metal content was determined by atomic absorption.

p426-ECA3 Yeast Expression Construct
AtECA3 from Arabidopsis 'Columbia-0' was cloned in two overlapping parts as initial attempts to amplify full-length cDNA failed. The 5# part was amplified from clone RZL01a05 (AV554735/AV545775) obtained from Kazusa DNA Research Institute (Asamizu et al., 2000) using the primers ECA3.F (5#-GCGGATCCAAACCCTTGGCTTTTCCAAC-3#) and ECA3-5#.R (5#-CAT-CCCATTGCTGTCAAAGACGGTACCTT-3#). The ECA3.F primer introduced a BamHI site, allowing the product to be cut with BamHI, and at an EcoRI site within the ECA3 sequence for ligation into the same sites of p426 (Mumberg et al., 1994) giving p426-5#-ECA3. The 3# part of ECA3 was amplified from reverse-transcribed RNA from leaf tissue using the primers ECA3-3#.F (5#-TGATGTCGGTGTCTAAGATATGTGT-3#) and ECA3.HR (5#-CGAAGCT-TCGAAAAAGCCATCTCTTGCT-3#) and cloned into pGEMTe. The primer ECA3-3#.F overlaps the end of the 5# part and the ECA3.HR primer introduced a HindIII site, allowing the ECA3 3# fragment to be excised with EcoRI and HindIII and ligated into the same sites of p426-5#-ECA3 to produce the full-length ECA3 cDNA in p426 (p426-ECA3). Sequencing of the final product confirmed that the full-length ECA3 was identical to the sequence predicted from genome sequencing. The GenBank accession number for ECA3 cloned in this study is EU082212.

YFP-ECA3 Construct
Gateway technology was used to produce the YFP-ECA3 construct. ECA3 cDNA was PCR amplified from the p426-ECA3 construct and integrated into the pENTR-TOPO vector by TOPO cloning (Invitrogen). ECA3 was then subcloned into the pEarleyGate 104 destination vector (Earley et al., 2006), which attaches YFP in-frame to the N terminus of ECA3.

ECA3 Promoter:GUS Construct
For tissue-specific expression studies, 1.5 kb of the 5#-untranslated region for ECA3 was amplified by high-fidelity PCR with primers 2548 (5#-CAC-CGTGTACAGTTGTATAGTTTACCAAC-3#) and 2549 (5#-GTTGGAAAAGC-CAAGGGTTTCTG-3#). The PCR product was cloned into pENTR/D-TOPO (Invitrogen) using Gateway technology and transferred to plasmid pMDC162 (Curtis and Grossniklaus, 2003) to generate a construct containing the open reading frame for a GUS reporter gene under the control of the ECA3 promoter region.
To compare the growth of K616 cells containing p426-ECA3 or p426 alone, cells were first grown in liquid culture overnight at 30°C in synthetic complete (SC) without uracil (pH 5.0) and containing 20 g L 21 Glc as the carbon source. Cultures were adjusted to the same OD 600 and aliquots were inoculated onto 2% (w/v) agar plates containing the SC 2 uracil medium (pH 4.9) with Gal or Glc and various additions as specified in ''Results.'' Plates were incubated at 30°C for 3 to 5 d.