cDNA microarray analysis of gene expression during Fe-deficiency stress in barley suggests that polar transport of vesicles is implicated in phytosiderophore secretion in Fe-deficient barley roots


*For correspondence (fax +81 3 6801 9557; email


To acquire Fe from soil, graminaceous plants secrete mugineic acid family phytosiderophores (MAs) from their roots. The secretion of MAs increases in response to Fe deficiency, and shows a distinct diurnal rhythm. We used a microarray that included 8987 cDNAs of rice EST clones to examine gene expression profiles in barley roots during Fe-deficiency stress. Approximately 200 clones were identified as Fe-deficiency-inducible genes, of which seven had been identified previously. In order to meet the increased demand for methionine to produce MAs, Fe-deficiency enhances the expression of genes that participate in methionine synthesis, as well as recycling methionine through the Yang cycle. Of these 200 genes, approximately 50 exhibited different transcription levels in Fe-deficient roots at noon and at night. Northern blot analysis of time course experiments confirmed that five of these genes exhibited a diurnal change in their level of expression. The diurnal changes in the expression of these genes suggest that polar vesicle transport is involved in the diurnal secretion of MAs.


Fe is an essential nutrient for plant growth and crop productivity. Although Fe is abundant in mineral soils (> 6%), under aerobic conditions in the physiological pH range, Fe is only sparingly soluble and not available to plants. Plants have therefore developed sophisticated and tightly regulated mechanisms for acquiring Fe from the soil. Graminaceous plants secrete Fe chelators, called mugineic acid family phytosiderophores (MAs), from their roots to solubilize Fe in the soil (Takagi, 1976). The resulting Fe3-MAs complexes are then reabsorbed into the roots through specific transporters in the cell membrane. The production and secretion of MAs markedly increases in response to Fe deficiency, and tolerance to Fe deficiency in graminaceous plants is strongly correlated with the quantity and quality of the MAs secreted. For example, rice, sorghum and maize secrete only deoxymugineic acid (DMA) in relatively low amounts and thus are susceptible to low Fe supply. In contrast, barley secretes large amounts of many kinds of MAs, including mugineic acid (MA), 3-hydroxymugineic acid (HMA) and 3-epi-hydroxymugineic acid (epi-HMA), and is therefore more tolerant to low Fe availability.

Three molecules of methionine are used for the synthesis of one MAs molecule (Figure 1) (Mori and Nishizawa, 1987; Shojima et al., 1989, 1990). We have isolated several genes from barley that encode enzymes in the biosynthetic pathway of MAs (Hvnas1–7, naatA, naatB, Ids2, Ids3) (Higuchi et al., 1999; Nakanishi et al., 1993; Okumura et al., 1994; Takahashi et al., 1999). Expression of all of these genes is dramatically enhanced by Fe deficiency resulting in increased secretion of MAs. It has been proposed that the Yang cycle functions to recycle the methionine required for continuous synthesis of MAs (Ma et al., 1995). In fact, genes encoding IDI1, FDH and APRT, enzymes closely related to the Yang cycle, are up-regulated in barley roots by Fe-deficiency stress (Itai et al., 2000; Suzuki et al., 1998; Yamaguchi et al. 2000a).

Figure 1.

The Yang cycle and the biosynthetic pathway of MAs in barley roots.

Methionine is activated to S-adenosylmethionine (SAM) by SAM synthetase as part of the Yang cycle, and three molecules of SAM are combined to form nicotianamine (NA) by nicotianamine synthase (NAS). The amino group of NA is transferred by nicotianamine aminotransferase (NAAT), and the resultant keto-form is reduced to deoxymugineic acid (DMA). The genes for the enzymes in boxes were previously isolated and shown to be Fe-deficiency-inducible genes.

In addition to the genes participating in the synthesis of MAs, several other genes have been identified as Fe3-deficiency-inducible genes in graminaceous plants. An Fe-MAs transporter gene, ys1, was isolated from maize, and steady-state levels of ys1 mRNA were increased by Fe starvation in maize roots (Curie et al., 2001). Other genes, such as a metallothionine-like gene (Ids 1)(Okumura et al., 1991), an eIF2Bα-like gene (IDI2) (Yamaguchi et al., 2000b), an ABC-type transporter gene (IDI7) (Yamaguchi et al., 2002) and an unknown gene (Ids 6) encoding a 36-kDa protein (Suzuki et al., 1997), were also isolated as Fe-deficiency-inducible in barley roots.

Previous studies elucidating the molecular mechanisms of tolerance to low Fe availability in graminaceous plants have focused primarily on the enzymes of the biosynthetic pathway and on isolation of the genes for MAs synthesis. However, other genes that may participate in plants' tolerance to low Fe availability still remain to be defined. For example, genes encoding factors that sense intracellular levels of Fe, transcriptional activators for regulating gene expression in response to Fe-deficiency, and components of signalling pathways to monitor Fe status in the environment, have not yet been identified. Furthermore, in contrast to the biosynthetic pathway of MAs, the molecular mechanism of MAs secretion still remains unclear. We have shown that MAs are secreted in the form of a monovalent anion via an anion channel (Sakaguchi et al., 1999). However, neither the genes nor the protein(s) responsible for transport of MAs to the outside of cells have been identified. MAs secretion in barley was reported to follow a distinct diurnal rhythm (Takagi et al., 1984). A secretion peak occurs just after initial illumination, and ceases within 2–3 h. In parallel with this diurnal secretion, there is a change in the shape of the vesicles in root cells of Fe-deficient barley. Because they have ribosomes on their cytoplasmic surface, these vesicles are thought to originate from the rough endoplasmic reticulum (rER). The vesicles stay swollen until the onset of MAs secretion, and become shrunken by the end of secretion (Nishizawa and Mori, 1987). It was proposed that these particular vesicles are the sites of MAs synthesis. We are interested in elucidating the sequence of events that links biosynthesis to the diurnal secretion of MAs.

Recently, microarray technology has emerged as a powerful tool for the analysis of gene expression. In the case of cDNA microarrays, gene fragments are fixed on a glass slide and hybridized with labelled target cDNA, prepared from RNA samples from different cells or tissue types, allowing direct and large-scale comparative analysis of gene expression. Several reports of the use of microarray analysis in plants have been published (Aharoni et al., 2000; Reymond et al., 2000; Ruan et al., 1998; Schena et al., 1995; Seki et al., 2001; Wang et al., 2000). In this study, we used microarray analysis to determine which genes are induced in response to Fe deficiency in barley. We used a cDNA microarray containing 8987 rice EST clones to analyse the gene expression profile in barley roots during Fe deficiency. Genes of cereal crops tend to be highly conserved at the DNA sequence level (Katrien and Mike, 2000), and this conservation allows the use of heterologous probes to identify orthologous DNA sequences in different species in DNA hybridization experiments. As a result, we were able to identify approximately 200 genes of which the expression levels were enhanced by Fe-deficiency.

Here, we show that the expression of genes that participate in methionine synthesis is enhanced in Fe-deficient barley roots. To meet the increased demands for methionine, methionine synthesis is enhanced in Fe-deficient barley, in addition to the Yang cycle for recycling methionine. Furthermore, we observed diurnal regulation of gene expression in the Fe-deficient roots. These findings suggest that polar transport of vesicles is involved in the diurnal secretion of MAs.


Fe-deficiency induced the expression of approximately 200 genes in barley roots

A microarray containing 8987 cDNAs of rice EST was used to analyse the gene expression profile during Fe-deficiency stress in barley roots. Since the rice genome is believed to contain approximately 30 000 genes, about one third was contained in this array. To detect differences in transcript levels between Fe-sufficient and Fe-deficient barley roots, mRNA was isolated from both Fe-deficient and Fe-sufficient barley roots at 11 am (Figure 2). Target cDNA was labelled with Cy5-dCTP and hybridized to the above microarray slides. The signal intensity from labelled targets derived from Fe-sufficient (+ Fe 11 am) and Fe-deficient (– Fe 11 am) roots was compared. On one slide, each cDNA clones was spotted in duplicate and we repeated the experiment twice. We regarded genes with an expression ratio greater than two-fold as Fe-deficiency-inducible genes. Approximately 200 genes were identified as Fe-deficiency inducible (Table 1).

Figure 2.

Strategy for identification of Fe-deficiency-inducible genes and genes related to the diurnal secretion of MAs.

Barley plants used in microarray experiments were grown hydroponically in a growth chamber with illumination from 9 : 30 am to 11 : 30 pm. At 11 am, plants were secreting large amounts of MAs, but at 1 am, plants were not secreting MAs. Fe-deficiency-inducible genes should therefore show differences in their expression between –Fe and +Fe at the same time (11 am). Furthermore, we compared the transcript levels of Fe-deficiency-inducible genes at 11 am with those at 1 am.

Table 1.  Fe-deficiency inducible genes
Element no.Accession no.Putative gene identificationRatio
Mas synthesis (3)
S-adenosylmethionine synthetase (Oryza sativa)
S-adenosylmethionine synthetase (Oryza sativa)
nicotianamine synthase 1 (Oryza sativa)
related to Met synthesis (3)
10-formyltetrahydrofolate synthetase (Spinacea oleracea)
Ca2+ sensitive 3′(2′),5-diphosphonucleoside 3′(2′) phophohydrolase (Oryza sativa)
Cysteine synthase (Oryza sativa)
Related to Yang cycle (2)
IDI1 (Hordeum vulgare)
formate dehydrogenase (Hordeum vulgare)
Related to cytoskeleton (5)
actin depolymerizing factor 6 (Arabidopsis thaliana)
tubulin alpha-1-chain (Oryza sativa)
actin depolymerizing factor (Zea mays)
alpha-tubulin (Oryza sativa)
alpha-tubulin (Oryza sativa)
Transcriptional regulation (5)
OSE3 luecine zipper protein (Oryza sativa)
zinc finger protein CONSTANTS like (Oryza sativa)
GF14-b protein (Oryza sativa)
GF14-c protein (Oryza sativa)
zinc-finger like protein (Arabidopsis thaliana)
Related to translation (4)
protein translation factor SUI1 homologue GOS2 (Oryza sativa)
translation initiation factor 4A2 eIF4A2 (Zea mays)
elongation factor 2 (Beta vulgaris)
IDI2 (Hordeum vulgare)
Others (87)
Ids 1 (Hordeum vulgare)
enolase (Oryza sativa)
enolase (Oryza sativa)
enolase (Oryza sativa)
Unknown (105)

Seven of these were recognised as Fe-deficiency-inducible genes in our previous studies (Table 2) (Mori, 1999). For example, nas encoding nicotianamine synthase (NAS), one of the critical enzymes of MAs biosynthetic pathway (Higuchi et al., 1999), was identified as having a high expression ratio. The gene encoding S-adenosylmethionine synthetase was also identified as Fe-deficiency inducible. We also identified the gene for formate dehydrogenase (FDH) (Suzuki et al., 1998) and the gene for the enzyme that catalyses the formation of 2-keto-methylthiobutyric acid (IDI 1) (Yamaguchi et al. 2000a). Both enzymes function in the Yang cycle to produce the methionine required for MAs synthesis.

Table 2.  Previously isolated Fe-deficiency inducible genes
Gene Present in this arrayDetected
nasHiguchi et al., 1999
naatTakahashi et al., 1999  
SMASTakizawa et al., 1996
Ids 1Okumura et al., 1991
Ids 2Okurmura et al., 1994  
Ids 3Nakanishi et al., 1993  
IDI1Yamaguchi et al., 2000
APRTItai et al., 2000  
FDHSuzuki et al., 1998
IDI2Yamaguchi et al., 2000
IDI7Yamaguchi et al., 2001  

To date, we have isolated 13 of the genes that are induced by Fe-deficiency in barley roots. Since rice produces only DMA and not other MAs, two of the 13 genes, Ids 2 and Ids 3, which encode enzymes that convert DMA to other MAs (Figure 1) (Kobayashi et al., 2001; Nakanishi et al., 2000) are not present in rice and therefore not expected to be detected by a rice cDNA microarray. In addition, Ids 6 and IDI7 may also not be present in rice; we were unable to find any rice ESTs homologous to these two genes in the database, and it is therefore reasonable that we did not detect these two genes using the rice cDNA microarray.

In contrast, the nicotianamine aminotransferase gene (naat) and the adenine phosphoribosyltransferase gene (APRT) should be present in the rice genome. However, although we found several rice ESTs homologous to these two genes, these ESTs were not included in the list of 8987 cDNAs on the microarray that we used. Presumably, the rice orthologs of naat and APRT were not contained in this microarray, and as a consequence naat and APRT were not detected as Fe-deficiency-inducible genes in this experiment. Otherwise, all of the genes known to be Fe-deficiency-inducible were detected by this microarray, confirming that this specific microarray analysis is reliable for identification of Fe-deficiency-inducible genes in barley roots.

Identification of genes with diurnally regulated transcript levels in Fe-deficient barley roots

In order to identify genes related to the diurnal secretion of MAs in Fe-deficient barley roots, further analysis was performed. RNA for target cDNA was prepared from Fe-deficient barley roots 1.5 h after the beginning of MAs secretion (– Fe 11 am) and when Fe-deficient barley roots were not secreting MAs (– Fe 1 am) (Figure 2). The signal intensities of labelled targets derived from Fe-deficient roots at 11 am (– Fe 11 am) and at 1 am (– Fe 1 am) were compared. We regarded Fe-deficiency-inducible genes with a ratio of signal intensify greater than two-fold as related to the diurnal secretion of MAs (Figure 2); approximately 50 such genes were identified (Table 3). To confirm the diurnal changes in transcript levels, a time course Northern blot analysis was performed with five of these genes (Figure 3). The transcript levels of each gene continuously increased until sunrise, and decreased after sunrise.

Table 3.  Genes related to MAs diurnal secretion
Element no.Accession no.Putative gene identificationRatio
Related to Mas synthesis (2)
D24418 S-adenosylmethionine synthetase (Oryza sativa)
nicotianamine synthase 1 (Oryza sativa)
Related to Yang cycle (1)
C28663 IDI1 (Hordeum vulgare)3.531
Possibly related to
transcriptional regulation (2)
strong similarity to fructokinase (Arabidopsis thaliana)
zinc-finger like protein (Arabidopsis thaliana)
Related to vacuole (3)
vacuolar proton-translocation ATPase subunit E Ylp (Hordeum vulgare)
delta-type tonoplast intrinsic protein (Triticum aestivum)
delta-type tonoplast intrinsic protein (Triticum aestivum)
Possibly related to
vesicle transport (3)
calmodulin 1 CaM1 (Oryza sativa)
N. tabacum mRNA for 22 kDa polypeptide (Nicotiana tabacum)
Ras-related protein RIC1 (Oryza sativa)
ADP-ribosylation factor 1 Os-ARF1 (Oryza sativa)
Others (22)
D41355 translation initiation factor 4A2 eIF4A2 (Zea mays)2.5691
Unknown (16)
Figure 3.

Diurnal changes in the transcript levels of Fe-deficiency-inducible genes related to secretion of MAs.

Five Fe-deficiency-inducible genes that show differences in their transcript levels between noon and night were used as probes.


In this study, we used microarray analysis to clarify the gene expression profile in barley roots during Fe-deficiency stress. Approximately 200 genes were identified as being Fe-deficiency inducible, including all of the genes that we have previously confirmed as Fe-deficiency inducible. This indicates that microarray analysis is an accurate and highly efficient method for differential screening of genes induced in plants under conditions of interest or stress. Almost all of the genes detected in this study are newly identified as Fe-deficiency-induced genes. This work also confirmed that microarray analysis is useful for obtaining very large amounts of gene expression data with a minimum number of experiments.

Methionine synthesis is enhanced in addition to methionine recycling in Fe-deficient barley roots

The Yang cycle functions to recycle the methionine required for continuous synthesis of MAs (Ma et al., 1995), and genes related to the Yang cycle are up-regulated in Fe-deficient barley roots. Furthermore, microarray analysis revealed that methionine synthesis itself is enhanced in Fe-deficient barley roots. The major pathway for methionine synthesis in plants is the transsulfuration pathway (Figure 4). In this pathway, the thiol group of cysteine is transferred to homoserine to produce homocysteine, via a cystathionine intermediate. Homocysteine is then converted to methionine by tetrahydrofolate-mediated methylation of homocysteine. The clone encoding cysteine synthase was detected as one of the genes induced by Fe-deficiency in barley roots. A clone encoding a Ca2+-sensitive 3′ (2′), 5′-diphosphonucleoside 3′ (2′)-phosphohydrolase (DPNPase) was also detected as an Fe-deficiency-inducible gene. DPNPase catalyses the conversion of adenosine 3′-phosphate 5′-phosphosulfate (PAPS) to adenosine 5′-phosphosulfate (APS) (Peng and Verma, 1995). APS provides S2, which is used with O-acetylserine in cysteine synthesis leading to methionine synthesis (Saito, 2000). It has been reported that cysteine synthase and serine acetyltransferase, which produce O-acetylserine, form a high molecular mass multienzyme complex (Saito et al., 1995). Through the formation of the multienzyme complex, the metabolic flow of intermediates from serine to cysteine can occur more efficiently, by preventing the diffusion of intermediary O-acetylserine. Therefore, it is conceivable that an increase in the transcript levels of cysteine synthase correlates to a simultaneous increase in the transcript level of serine acetyltransferase. In addition, a gene homologous to 10-formyltetrahydrofolate synthase (SYN) was also detected as one of the genes whose expression was induced by Fe-deficient treatment. SYN is an enzyme in the pathway for tetrahydrofolate production, which mediates methylation of homocysteine to produce methionine (Chen et al., 1997). Thus, it is clear that the expression of genes encoding enzymes participating in methionine synthesis is increased in Fe-deficient barley roots. This suggests that in Fe-deficient barley roots the transsulfuration pathway for methionine synthesis through homocysteine is enhanced in addition to the action of the Yang cycle in recycling methionine, in order to meet the increased demand for precursor methionine in MAs synthesis (Figure 4).

Figure 4.

Methionine synthesis through the transsulfuration pathway.

The genes for the enzymes in boxes were identified as Fe-deficiency-inducible clones by the microarray experiment.

Genes involved in transcription were induced by Fe-deficiency

14-3-3 protein genes (homologous to OsGF14b and OsGF14c) were detected as Fe-deficiency inducible. These proteins form a conserved eukaryotic family of regulatory proteins that function via phosphorylation–dependent interactions with a wide range of target proteins (Roberts, 2000). One of the first groups of 14-3-3 proteins to be identified in plants was discovered during studies of transcription. 14-3-3 proteins that interact with G-box-binding complexes were purified from maize and Arabidopsis, and termed G-box factor 14-3-3 (GF 14) proteins (de Vetten et al., 1992; Lu et al., 1992). A potential role for 14-3-3 proteins in more general regulation of transcription was also demonstrated.

Fe-deficiency induced two genes for zinc finger proteins, a CONSTANS family zinc finger protein and homologues of the Arabidopsis zinc finger protein. In addition, the gene for a leucine zipper protein (OSE3) was also induced. Until now, neither an Fe-deficiency-responsive element in the promoter region of Fe-deficiency-inducible genes, nor a trans factor that interacts with this element, has been described in plants. Therefore, these genes are promising candidates for transcriptional activators regulating transcription in response to Fe-deficiency.

Genes participating in translation were also induced by Fe-deficiency

Genes for factors that regulate mRNA translation were also found in Fe-deficiency-inducible genes. Fe-deficiency induced the gene for GOS2 protein and homologues of maize eIF4A2, which are involved in initiation of translation. We previously isolated IDI2 (Iron Deficiency Induced gene 2) – a homologue of an eIF2Bα-like gene in Arabidopsis (Yamaguchi et al., 2000b). Since eIF2Bα-like protein seems to be involved in initiation of translation, IDI2 might play a similar role in initiation of translation. A homologue of EF2, an elongation factor in Beta vulgaris, was detected as an Fe-deficiency-inducible gene. These results suggest that protein synthesis in barley roots in response to Fe-deficiency is regulated not only at the transcriptional level but also at the translational level.

Many of the Fe-deficiency-inducible genes detected in this experiment remain of unknown function. Further work is needed to elucidate the role of these genes in plants' tolerance to low Fe availability.

Genes possibly related to the diurnal secretion of MAs

The molecular mechanism of MAs secretion has not been elucidated to date. We performed further microarray analyses to identify genes that play a role in the diurnal secretion of MAs in barley. As a result, approximately 50 Fe-deficiency-inducible genes were obtained as candidates that are possibly related to the diurnal secretion of MAs. Some of the genes that function in MAs synthesis (SAMS, nas and IDI1) were included within the 50 genes detected. This suggests that MAs' synthesis is at least partly controlled by diurnal rhythms.

Polar vesicle transport may be implicated in the diurnal secretion of MAs

Northern blot analysis was performed using the genes for calmodulin, translation initiation factor 4A2, ras-related small GTP-binding protein (GTPase), ADP-ribosylation factor 1 (ARF1) and a gene homologous to Nicotiana tabacum mRNA for the 22-kDa polypeptide (Ganet et al., 1996), to confirm changes in transcription levels. Transcript levels for all these genes increased before sunrise and gradually decreased after sunrise. These diurnal changes in transcript levels support the idea that these genes are involved in the diurnal secretion of MAs. It is of great interest that homologues to both Osric1 and OsARF1 were identified as genes related to the diurnal secretion of MAs. Osric1 is a member of the family of Rab-1-related genes. Current information suggests that Rab proteins are localized to the cytoplasmic face of all organelles involved in intracellular transport, and regulate the specificity and directionality of vesicular transport from a source to a target compartment in many organisms. In plants, it has been reported that a Rab1 GTPase is required for transport between the ER and the Golgi apparatus in Arabidopsis (Batoko et al. 2000). Similarly, ARFs, ubiquitous GTP-binding proteins, are required for maintaining the integrity of organelle structure and intracellular transport in many organisms (Moss and Vaughan, 1998). In Arabidopsis, Steinmann et al. (1999) revealed that a membrane–associated guanine nucleotide exchange factor (GEF) on ARF (ARF GEF) is essential for co-ordinated polar localization of the auxin efflux carrier AtPIN 1 in embryogenesis.

Targeting of transmembrane proteins like AtPIN 1 depends largely on sorting signals residing within the cytosolic tails of the proteins, which are recognised by specific receptors. In mammalian cells, di-leucine (LL; where the leucines can also be replaced by isoleucines) and YXXϕ (where Y refers to tyrosine, X refers to any amino acid residues, and ϕ refers to hydrophobic residues with a bulky side chain) motifs are among the best understood. Since AtPIN 1 contains multiple YXXϕ and LL signals, these motifs may function in ARF GEF-mediated polar localization of AtPIN 1 (Zuo et al., 2000).

We have proposed that MAs are synthesized in the vesicles that appear in the cortex cells of Fe-deficient barley roots. NAS and NAAT, both enzymes critical to MAs synthesis, are targeted to these rER-derived vesicles (unpublished). NAS is a type-I transmembrane protein (Higuchi et al., 1999), and presumably resides in the membrane of these rER-derived vesicles. Interestingly, HvNAS1 contains two putative polarized targeting motifs within its cytosolic tail (tyrosine 105 to leucine 108 and leucine 115, 116), and these motifs are completely conserved in all NAS genes isolated so far from barley, rice, maize, Arabidopsis and tomato. This indicates that polar transport of these vesicles is involved in the secretion of MAs. In addition, these vesicles were observed to accumulate in the epidermal cells at the cell periphery facing the rhizosphere just before sunrise (Figure 5). Therefore, we hypothesize that MAs synthesized in the rER-derived vesicles are localized to the cell boundaries facing the rhizosphere by a polar vesicle transport process, in which ARF and Rab1 GTPase are components, leading to the diurnal secretion of MAs.

Figure 5.

Electronmicrograph of Fe-deficient barley root epidermal cells.

5A: rER-derived vesicles (arrow heads) accumulated at the cell periphery facing the rhizosphere. Magnification is × 12 000; bar = 1 µm.

5B: Higher magnification (x 30 000) of 5A bar = 1 µm.CW indicates cell wall; G indicates Golgi apparatus; M indicates mitochondria; V indicates vacuole; rER indicates rough endoplasmic reticulum.

In this scenario, diurnally regulated transcription of calmodulin may regulate vesicle transport through the interactions of kinesin-like calmodulin binding protein and cytoskeletal microtubules (Vos et al., 2000). Alternatively, calmodulin may regulate the K+ release that occurs simultaneously with MAs secretion. We have reported that MAs are secreted in the form of a monovalent anion via an anion channel using the K+ gradient between the cytoplasm and the cell exterior (Sakaguchi et al., 1999). As a result, MAs are secreted with equimolar potassium (Takagi et al., 1984). KCO1, which is an outward rectifying K+ channel of Arabidopsis thaliana, is strongly dependent on Ca2+ (Czempinski et al., 1997). An unidentified K+ channel participating in K+ secretion in parallel with MAs secretion in barley roots might have properties similar to KCO1, and be regulated by the calmodulin-mediated signalling pathway.

It must be pointed out that a homologue of the Arabidopsis fructokinase gene was included in the 50 genes identified as possibly being related to the diurnal secretion of MAs. Recent thought implicates fructokinase as a sugar sensor that can relay the initial signal to downstream cascade components in plants, yeast and mammals (Pego and Smeekens, 2000). Sugar photosynthesized in the daytime and translocated to the roots may be a component of the signal system for the diurnal expression of genes involved in the secretion of MAs.

Using a rice cDNA microarray, we demonstrated that methionine synthesis through the transsulfuration pathway is enhanced in Fe-deficient barley roots, in addition to methionine recycling via the Yang cycle. We also identified five genes with diurnally regulated transcript levels in Fe-deficient barley roots. The diurnal expression of these genes indicates that polar vesicle transport is involved in the diurnal secretion of MAs.

Materials and methods

Plant materials and growth conditions

Barley seeds (Hordeum vulgare L. cv. Ehimehadaka no. 1) were germinated and the seedlings were grown hydroponically, as described previously (Mori and Nishizawa, 1987). For Fe-deficiency treatment, plants were transferred to culture solution without Fe when the third leaf emerged. After 2-weeks of treatment, the roots of plants were harvested when secretion of MAs ceased (1 am) and 1.5 h after MAs secretion began (11 am). For Northern blot analysis, the roots of 2-week-old-treated plants were harvested when secretion of MAs stopped (1 am), 3.5 h (6 am) and 30 min (9 am) before secretion of MAs began and 1.5 h (11 am), 3.5 h (1 pm), and 5.5 h (3 pm) after MAs secretion began, respectively. Plants growing in a culture solution with Fe for 2 weeks were harvested 1.5 h after the start of MAs secretion (11 am) as Fe-sufficient plants.

For electron microscopy, root of Fe-deficient barley before MAs secretion was quickly frozen by high pressure freezing method using HPM 010 (BAL-TEC, Liechtenstein) and freeze-substituted as previously reported (Nishizawa and Mori, 1989).

RNA preparation and Northern blot analysis

Total RNA was extracted from barley roots and 40 µg was used for microarray analysis. For Northern blot analysis, 20 µg of total RNA was used. Radiolabelled DNA probes for each gene were prepared using individual cDNA clones of the rice cDNA microarray. Northern blots were analysed using BAS 3000 (FujiFilm).

Microarray preparation

Microarray analysis was performed basically as previously described (Yazaki et al., 2000). The rice cDNA microarrays were prepared on aluminum-coated and DMSO-optimized glass slides. The sequences used in the construction were generated by PCR. The PCR products were purified by QIAquick 96-column (QIAGEN). DNA solutions were arrayed by robotics, Array Spotter Generation III.

Fluorescent labelling of probe

Isolated total RNA was reverse-transcribed with Cy5 (Amersham Pharmacia). Reactions were incubated for 2.5 h at 42°C with 80 µg of total RNA, oligo(dT)25, random nonamer, control cRNA, 1XSSII reaction buffer, DTT, 2 mm dATP, 2 mm dTTP, 2 mm dGTP,1 mm dCTP, 1 mm Cy5dCTP and SSII reverse transcriptase. The reactions were denatured at 94°C for 3 min and the RNA was degraded by incubation with 2 µl of 2 N NaOH at 37°C for 15 min Following degradation, the mixture was neutralized with 10 µl of 2 m HEPES buffer. The labelled probes were purified using a Qiaquick PCR Purificaiton Kit (QIAGEN) and dried using a vacuum concentrator. The dried probes were resuspended in 6 µl of water and denatured at 95°C for 4 min; 1.5 µl of Oligo A80 (1 mg ml−1) in 7.5 µl of 4× hybridization buffer containing SSC, SDS, Denhardt's solution, salmon sperm DNA, and 15 µl of formamide were added to the resuspended probes. A final volume of 30 µl was used for hybridization.

Hybridization on microarrays and analysis

After hybridization, the glass slides were washed in 1× SSC/0.2% SDS for 10 min at 55°C in the dark, then in 0.1× SSC/0.2% SDS for 10 min at 55°C twice in the dark, and finally in 0.1× SSC for 1 min at room temperature twice. After the final wash, the slides were briefly rinsed with distilled water and air-dried. The hybridized and washed microarrays were scanned using an Array Scanner Generation III (Amersham Pharmacia). ArrayGauge (FujiFilm) was used for image analysis.