The molecular mechanisms of plant responses to iron (Fe) deficiency remain largely unknown. To identify the cis-acting elements responsible for Fe-deficiency-inducible expression in higher plants, the barley IDS2 (iron deficiency specific clone no. 2) gene promoter was analyzed using a transgenic tobacco system. Deletion analysis revealed that the sequence between −272 and −91 from the translational start site (−272/−91) was both sufficient and necessary for specific expression in tobacco roots. Further deletion and linker-scanning analysis of this region clearly identified two cis-acting elements: iron-deficiency-responsive element 1 (IDE1) at −153/−136 (ATCAAGCATGCTTCTTGC) and IDE2 at −262/−236 (TTGAACGGCAAGTTTCACGCTGTCACT). The co-existence of IDE1 and IDE2 was essential for specific expression when the −46/+8 region (relative to the transcriptional start site) of the CaMV 35S promoter was used as a minimal promoter. Expression occurred mainly in the root pericycle, endodermis, and cortex. When the −90/+8 region of the CaMV 35S promoter was fused, the −272/−227 region, which consists of IDE2 and an additional 19 bp, could drive Fe-deficiency-inducible expression without IDE1 throughout almost the entire root. The principal modules of IDE1 and IDE2 were homologous. Sequences homologous to IDE1 were also found in many other Fe-deficiency-inducible promoters, including: nicotianamine aminotransferase (HvNAAT)-A, HvNAAT-B, nicotianamine synthase (HvNAS1), HvIDS3, OsNAS1, OsNAS2, OsIRT1, AtIRT1, and AtFRO2, suggesting the conservation of cis-acting elements in various genes and species. The identification of novel cis-acting elements, IDE1 and IDE2, will provide powerful tools to clarify the molecular mechanisms regulating Fe homeostasis in higher plants.
Iron (Fe) is an essential nutrient required for various cellular events, including respiration, chlorophyll biosynthesis, and photosynthetic electron transfer, and is also a component of the Fe-S cluster present in numerous enzymes. Although abundant in mineral soils, Fe is sparingly soluble under aerobic conditions at high soil pH. Consequently, plants grown on calcareous soils often exhibit severe chlorosis (lime-induced chlorosis) because of Fe deficiency, which is a major agricultural problem resulting in reduced crop yields (Marschner, 1995). To acquire Fe, higher plants have evolved two major strategies (Marschner et al., 1986). Non-graminaceous plants enhance proton excretion into the rhizosphere to lower pH, induce the expression of ferric-chelate reductase to reduce Fe to the more soluble ferrous form at the root surface, and transport the ferrous ions generated across the root plasma membrane. This is referred to as Strategy I. Graminaceous plants have another strategy that is referred to as Strategy II, which is mediated by natural Fe chelators, the mugineic acid family phytosiderophores (MAs). Graminaceous plants synthesize and secrete MAs from their roots to solubilize Fe(III) in the rhizosphere (Takagi, 1976). The resulting Fe(III)–MA complexes are taken up into the root by a specific transporter in the plasma membrane (Mihashi and Mori, 1989; Takagi, 1976). To date, seven MAs and their biosynthetic pathways have been identified (Ma et al., 1999; Mori and Nishizawa, 1987; Nomoto et al., 1987; Shojima et al., 1990).
Many genes involved in each strategy have been cloned. The crucial genes in Strategy I have been isolated from Arabidopsis thaliana: FRO2 (a ferric-chelate reductase; Robinson et al., 1999) and IRT1 (a ferrous Fe transporter; Eide et al., 1996). FRO2 expression is regulated at the transcriptional level in response to Fe deficiency. In contrast, Fe controls IRT1 expression at the transcript and protein accumulation levels (Connolly et al., 2002). Most of the barley Strategy-II genes have been cloned. Of these, S-adenosylmethionine synthetase (HvSAMS; Takizawa et al., 1996), nicotianamine synthase (HvNAS, Higuchi et al., 1999; nashor, Herbik et al., 1999), and nicotianamine aminotransferase (HvNAAT; Takahashi et al., 1999) genes encode enzymes for the biosynthesis of 2′-deoxymugineic acid, the first member of MAs in the biosynthetic pathway. Two genes isolated using the differential hybridization method, iron deficiency specific clone no. 2 (IDS2) (Okumura et al., 1994) and IDS3 (Nakanishi et al., 1993), encode dioxygenases that hydroxylate the C-3 and C-2′ positions of MAs, respectively (Kobayashi et al., 2001; Nakanishi et al., 2000). Some genes involved in recycling methionine, which is the precursor of MAs (Mori and Nishizawa, 1987), have also been isolated: IDI1 (Yamaguchi et al., 2000a), the formate dehydrogenase gene HvFDH (Suzuki et al., 1998), and the adenine phosphoribosyltransferase gene HvAPT1 (Itai et al., 2000). The expression of these genes is induced by Fe deficiency in roots (Negishi et al., 2002). Recently, a gene encoding a possible Fe(III)-MA transporter, Yellow Stripe1 (YS1), was cloned from maize (Curie et al., 2001). Under Fe deficiency, the YS1 mRNA level is increased in both roots and shoots. In addition to the genes involved in MA production and transport, we have isolated other Fe-deficiency-inducible genes from barley, including: a metalothionein-like gene, IDS1 (Okumura et al., 1991); an eIF2Bα-like gene, IDI2 (Yamaguchi et al., 2000b); and a tonoplast located ABC-type transporter gene, IDI7 (Yamaguchi et al., 2002). All of the Fe-deficiency-inducible genes isolated from barley are expressed almost exclusively in roots. It is possible that the root-specific expression of these genes is important for Fe acquisition from the rhizosphere. Furthermore, recently developed microarray techniques make it possible to search simultaneously for numerous genes induced by Fe deficiency. For instance, many of the genes involved in MA synthesis, along with some genes possibly involved in the diurnal secretion of MAs, have been confirmed to be induced by Fe deficiency in barley roots (Negishi et al., 2002). Microarray analysis in Arabidopsis revealed that Fe deficiency increases aerobic respiration in shoots and anaerobic respiration in roots (Thimm et al., 2001). In tomato roots, several regulatory, signaling, and transporter genes have been shown to be similarly induced by phosphorus, potassium, and Fe deficiencies (Wang et al., 2002).
Despite the number of genes isolated, little is known about the regulation of gene expression in response to Fe deficiency. Especially, no Fe-deficiency-responsive elements have been identified in the promoter regions of Fe-deficiency-inducible genes. To clarify the expression mechanism of Fe-deficiency-inducible barley genes, we introduced the promoter regions of such genes into other species and analyzed the heterologous expression. Transgenic rice plants carrying a barley genomic clone containing HvNAS1, IDS3, or HvNAAT-A and HvNAAT-B expressed their encoding gene(s) in an Fe-deficiency-inducible manner (Higuchi et al., 2001a; Kobayashi et al., 2001; Takahashi et al., 2001). Moreover, transgenic tobacco plants carrying either HvNAS1 or the barley IDS2 promoter connected to the GUS gene also strongly expressed GUS in response to Fe deficiency in their roots (Higuchi et al., 2001b; Yoshihara et al., 2003), suggesting conservation of the mechanisms regulating these genes in plants that use two different Fe-acquisition strategies: barley (Strategy II) and tobacco (Strategy I). The IDS2 promoter also drives expression under conditions of manganese and zinc deficiency in roots of native barley (Okumura et al., 1994), whereas it does not respond to manganese or zinc deficiency in transgenic tobacco (Kobayashi et al., 2003). This suggests differences between barley and tobacco in the endogenous trans-acting factors that sense manganese and zinc deficiencies. 5′-Deletion analysis of the HvNAS1 and IDS2 promoters suggested the presence of dominant cis-acting elements that confer Fe-deficiency-inducible, root-specific expression within −348 of HvNAS1 and −272 of IDS2 (relative to the translational start site; Higuchi et al., 2001b; Yoshihara et al., 2003). Although these reports included important information on the expression mechanism of Fe-deficiency-inducible genes, the information was not sufficient for identification of the cis-acting elements. In the present study, 3′-deletion and mutation series of the IDS2 promoter were generated and the Fe-deficiency response was analyzed in a transgenic tobacco system. We identified two novel cis-acting elements, iron-deficiency-responsive element 1 (IDE1) and IDE2, which synergistically induced Fe-deficiency-specific expression in tobacco roots. Sequence comparison with other Fe-deficiency-inducible promoters revealed the possible conservation of cis-acting elements among various plant genes and species.
The sequence between −272 and −91 of the IDS2 promoter is essential for Fe-deficiency-inducible expression in tobacco roots
Previously, we reported an analysis of eight 5′-deletion clones between positions −1696 and −47 of the IDS2 promoter in tobacco (Yoshihara et al., 2003). To further investigate the function of this promoter region, we designed eight new 3′-deletion clones (clones 1-46, 2-46, 3-46, 4-46, 5-46, 6-46, 7-46, and 8-46; Figure 1). The 3′-end of the longest promoter fragment (−1696/−91) was situated at the 5′-border of the putative TATA box (Okumura et al., 1994). Each fragment of the IDS2 promoter was fused to the CaMV 35S minimal promoter (35SΔ46; −46/+8 from the transcriptional start site)-GUS construct, and transferred into tobacco plants. The response to Fe deficiency was determined by measuring the GUS activity in T1 transgenic tobacco plants. Each clone carrying the IDS2 promoter expressed more than the background level of GUS activity in roots under both Fe deficiency and sufficiency (Figure 1). However, prominent GUS expression was observed only in Fe-deficient roots of clone 1-46, which carried the longest fragment. The relative ratio of GUS activity in Fe-deficient and Fe-sufficient roots (induction ratio) increased up to 196-fold. The other seven clones (clones 2-46, 3-46, 4-46, 5-46, 6-46, 7-46, and 8-46), which lacked the sequence from −272 to −91, showed no induction in response to Fe deficiency, indicating the importance of the −272/−91 sequence for Fe-deficiency-inducible expression in roots. The deletion from −1129 to −1274 reduced the expression in Fe-sufficient roots (clone 6-46 versus clone 7-46), while the other deletions did not significantly affect the expression in Fe-sufficient roots. In leaves, clones 1-46 and 2-46 expressed more than the background level of GUS activity only under Fe-deficiency. The other six clones (clones 3-46, 4-46, 5-46, 6-46, 7-46, and 8-46) showed little or no substantial expression in leaves. Clone 0-46, which lacked the IDS2 promoter, expressed only background levels of GUS in roots and leaves.
The sequence between −272 and −91 contains two regions that respond to Fe deficiency
Four 3′-deletion fragments within the −272/−91 sequence were produced and fused to 35SΔ46 (clones 11-46, 12-46, 13-46, and 14-46) or the −90/+8 region of the CaMV 35S promoter (35SΔ90; clones 11-90, 12-90, 13-90, and 14-90; Figure 2). While 35SΔ46 contains only the basic apparatus for transcription (i.e. the TATA box), 35SΔ90 originally contained the sequence for preferential expression in roots when introduced into tobacco (Benfey et al., 1989). Clone 11-46, carrying the −272/−91 sequence fused to 35SΔ46, conferred strong GUS expression in roots only under conditions of Fe deficiency (Figure 2). Although its level of expression was only about one-seventh that of the 5′-extended derivative (clone 1-46 in Figure 1), the induction ratio of GUS activity with Fe deficiency was roughly the same (about 200-fold). Marked induction by Fe deficiency was also observed in the roots of clone 12-46. The decrease in GUS activity in Fe-deficient roots of clone 12-46 compared with clone 11-46 was not statistically significant. Further deletion from −136 to −181 markedly reduced the expression in Fe-deficient roots. The level of expression of clone 13-46 in Fe-deficient roots was significantly lower than that of clone 12-46 but significantly higher than that of clone 0-46. The deletion from −181 to −227 had little effect, although clone 14-46 showed no significant induction of GUS activity in Fe-deficient roots as compared with clone 0-46. All four IDS2 clones fused to 35SΔ46 showed little or no expression in Fe-sufficient roots, or in Fe-sufficient and Fe-deficient leaves.
When the IDS2 promoter fragments were fused to 35SΔ90 instead of 35SΔ46, all four clones clearly induced GUS activity in roots in response to Fe deficiency (Figure 2; clones 11-90, 12-90, 13-90, and 14-90). Deletions of the IDS2 promoter between −91 and −227 had no pronounced effect, except a statistically insignificant reduction in activity in Fe-deficient roots because of the deletion from −136 to −181. Of note, clones 11-90, 12-90, 13-90, and 14-90 also showed greater than background GUS activity in Fe-deficient leaves. The GUS activity in Fe-deficient leaves of clones 11-90, 12-90, and 13-90 was significantly higher than that of clone 0-90, which contained 35SΔ90 but lacked the IDS2 promoter. Clones 11-90 and 14-90 expressed significantly higher GUS activity under Fe deficiency than under Fe sufficiency, while clones 12-90 and 13-90 did not. The GUS activities of clones 11-90, 12-90, 13-90, 14-90, and 0-90 in Fe-sufficient roots were comparable to each other, and were significantly higher than the wild-type level. Clone 0-90 did not respond to Fe deficiency.
These 3′-deletion analyses suggested the presence of two regions containing cis-acting elements responsive to Fe deficiency. The −180/−136 region appeared essential for conferring a strong response to Fe deficiency in roots when fused to 35SΔ46, whereas the −272/−227 region had no clear function when fused to 35SΔ46, but conferred a distinct response to Fe deficiency in roots when fused to 35SΔ90.
The two cis-acting elements between −272 and −91, which synergistically confer Fe-deficiency-inducible, root-specific expression, are situated at −153/−136 and −262/−236
The site of the cis-acting element within the −180/−136 region was determined by linker-scanning analysis. Five clones harboring a 9-bp mutation within the −180/−136 region were constructed and individually connected downstream of the −272/−181 sequence (Figure 3a; from clones M1-46 to M5-46). Native fragments of −180/−136 with or without the upstream fragment of −272/−181 or −272/−227 were also constructed (clones N1-46, N2-46, and N3-46). The −272/−181 and −272/−227 fragments were modified to contain a 2-bp substitution and 1-bp insertion at the 3′ junction to connect to the −180/−136 region. Each fragment was fused to the 35SΔ46-GUS construct and introduced into tobacco. Some of the clones expressed distinct GUS activity in roots, specifically responding to Fe deficiency (Figure 3b). No substantial expression was observed in leaves (data not shown). Clone N1-46 induced the same level of GUS expression as clone 12-46, confirming that the mutation at −182/−181 does not influence function. Three mutations within the sequence between positions −180 and −154 (clones M1-46, M2-46, and M3-46) did not significantly alter the specificity. In contrast, two mutations within the sequence between positions −153 and −136 (clones M4-46 and M5-46) markedly reduced the responsiveness to Fe deficiency, demonstrating that the cis-acting element within the −180/−136 region is situated at −153/−136. An internal deletion of the sequence between −226 and −181 (clone N2-46) did not reduce the responsiveness to Fe deficiency, whereas the 5′-deletion of the whole distal fragment (clone N3-46) completely eliminated GUS expression. Every derivative of clones M1-46 to M5-46 that lacked the −272/−181 region lacked GUS expression (data not shown). This observation indicates that an element(s) within the −272/−227 region is (are) essential for the function of the element at −153/−136, at least when 35SΔ46 is used as a minimal promoter.
The site of the element within the −272/−227 region was determined similarly using clone N2-46 and mutated derivatives. Five clones harboring 9- to 10-bp mutations within the −272/−227 region were constructed and individually connected upstream of the −180/−136 sequence (Figure 4a; from clones M21-46 to M25-46). Each fragment was fused to the 35SΔ46-GUS construct and introduced into tobacco. No substantial expression was observed in the leaves of these clones (data not shown). Clones M21-46 and M25-46 expressed levels of GUS in roots in response to Fe deficiency, similar to clone N2-46 (Figure 4b). Conversely, clones with mutations within the sequence between positions −262 and −236 (clones M22-46, M23-46, and M24-46) showed markedly reduced responsiveness to Fe deficiency as compared to clone N2-46, localizing the cis-acting element to −262/−236. In conclusion, linker-scanning analysis indicated that two cis-acting elements situated at −262/−236 and −153/−136 drove Fe-deficiency-inducible, root-specific expression synergistically when connected to the 35SΔ46 minimal promoter.
Spatial pattern of Fe deficiency-induced expression because of IDS2 promoter fragments
The tissue specificity of expression was confirmed by histochemical staining (Figure 5). Of the clones carrying the IDS2 promoter fused to 35SΔ46, all of the Fe-deficiency-responsive fragments from the 3′-deletion or mutation analyses (Figures 1–4) showed similar expression patterns in Fe-deficient roots (Figure 5a–c). Root cross-sections were stained mainly in the pericycle, endodermis, and cortex (Figure 5a,b). Weaker expression was observed in the epidermis. Expression was strongest near apical root zones exhibiting typical Fe-deficiency phenotypes, but absent in the root cap in most cases (Figure 5c). However, marked expression was also seen in the epidermis and root cap in some transformant lines (data not shown). No substantial expression was observed in leaves or Fe-sufficient roots in any clone, with the exception of clone 1–46, which showed weak expression in Fe-deficient leaf veins and Fe-sufficient root cortex (data not shown).
In contrast, 3′-deletion clones carrying 35SΔ90 expressed GUS in both Fe-sufficient and Fe-deficient roots, with much stronger activity under conditions of Fe deficiency. Strong expression was observed virtually throughout Fe-deficient roots (Figure 5d), although the apical zone was sometimes unstained. Root cross-sections revealed especially strong expression in the pericycle, endodermis, and epidermis (Figure 5d). Under conditions of Fe sufficiency, moderate expression was observed in the endodermis, cortex, and stele, but not in the epidermis (data not shown). Faint expression was sometimes observed in the veins of Fe-sufficient and Fe-deficient leaves (data not shown). These spatial patterns of expression were the same in the four 3′-deletion clones. The expression pattern in clone 0-90 lacking the IDS2 fragment, under conditions of both Fe sufficiency and deficiency, was similar to that in Fe-sufficient transformants carrying the IDS2 fragments fused to 35SΔ90 (data not shown). These results suggest that the IDS2 promoter fragments and 35SΔ90 synergistically drive expression in Fe-deficient roots in spatially different patterns from the expression conferred by either the IDS2 promoter fragments or 35SΔ90 alone.
Identification of synergistically functioning cis-acting elements in the IDS2 promoter
We analyzed the Fe-deficiency-inducible, root-specific expression of the barley IDS2 gene using a stable transformation system with transgenic tobacco. A stable transformation system was thought to be more favorable than transient assay systems for analyzing the properties of the IDS2 gene promoter, including tissue-specific expression and the response to nutrient deficiencies. The IDS2 promoter was previously confirmed to be responsive to Fe deficiency in tobacco, and 5′-deletion analysis demonstrated that the sequence between −272 and −47 contains the cis-acting elements required for Fe-deficiency-inducible expression (Yoshihara et al., 2003). The results of our 3′-deletion analysis (Figure 1) clearly demonstrated that the −272/−91 sequence was not only sufficient, but also essential for the specific response to Fe deficiency in roots. Considering the results of 5′-deletion analysis (Yoshihara et al., 2003), the upstream sequence between −1696 and −273 was thought to contribute to Fe-deficiency-inducible expression in roots only in the presence of the proximal −272/−91 sequence.
Further analysis showed that two novel cis-acting elements in the −272/−91 sequence synergistically direct expression in response to Fe deficiency in roots. When 35SΔ46 was used as a minimal promoter, the 3′-deletion from −136 to −181 markedly reduced the Fe-deficiency-inducible expression (Figures 2 and 6; clone 12-46 versus clone 13-46). This strongly suggested the presence of at least one cis-acting element in the −180/−136 region. Linker-scanning analysis revealed that mutation at either −153/−145 or −144/−136 severely reduced the responsiveness to Fe deficiency (Figures 3 and 6; clones M4-46 and M5-46). Therefore, the presence of a cis-acting element at −153/−136, which we designated IDE1 (Figure 6), was clearly demonstrated. However, the presence of IDE1 was not sufficient to confer the response to Fe deficiency when the sequence upstream of −181 was deleted (Figures 3 and 6; clone N3-46). Conversely, deletion of IDE1 did not markedly reduce the response to Fe deficiency in the presence of 35SΔ90 (Figures 2 and 6; clone 12-90 versus clones 13–90 and 14–90); the −272/−227 region still responded to Fe deficiency when fused to 35SΔ90 (Figures 2 and 6; clone 14-90). This suggested the presence of other cis-acting elements in the −272/−227 region. Linker-scanning analysis using 35SΔ46 as a minimal promoter revealed that mutation at −262/−254, −253/−245, or −244/−236 resulted in almost complete loss of the response to Fe deficiency (Figures 4 and 6; clones M22-46, M23-46, and M24-46). This demonstrated that the cis-acting element, which we designated IDE2 (Figure 6), resides in −262/−236.
IDE1 and IDE2 are expected to show the synergistic relationship outlined in Figure 6. When 35SΔ46 was used as a minimal promoter, deletion or mutation of either IDE1 or IDE2 resulted in marked, if not complete, loss of responsiveness to Fe deficiency (clones 13-46, M4-46, M5-46, N3-46, M22-46, M23-46, and M24-46). Therefore, IDE1 and IDE2 function synergistically; i.e. trans-acting factors that bind to the elements synergistically drive the Fe-deficiency-inducible expression by affecting the basic transcription apparatus. Analysis of the −272/−91 region of the IDS2 promoter indicated that this Fe-deficiency responsiveness was consistently accompanied by root-specific expression. The deletions and mutations did not alter the tissue specificity of the expression in roots (Figure 5a–c). Therefore, the pair of IDE1 and IDE2 confers tissue specificity as well as responsiveness to Fe deficiency. The distance between IDE1 and IDE2 in the native IDS2 promoter was not essential for their function, because the internal deletion of −226 to −181 did not reduce the responsiveness (Figure 3; clone N1-46 versus clone N2-46). When 35SΔ90 was fused, the Fe deficiency responsiveness conferred by the −272/−227 region, which consists of IDE2 and an additional 19 bp, was obvious in both the presence (clones 11-90 and 12-90) and the absence (clones 13-90 and 14-90) of IDE1. It would be of interest to confirm whether IDE1 connected to 35SΔ90 confers responsiveness in the absence of IDE2. Synergistic control of the expression of more than one element has been observed in other plant promoters, such as the light-inducible, green-organ-specific expression of rbcS gene promoters (Benfey and Chua, 1989; Donald and Cashmore, 1990), floral expression of chalcone synthase gene promoters (Faktor et al., 1996), endosperm-specific expression of a wheat glutenin gene (Albani et al., 1997), and sugar-repression of a rice α-amylase gene promoter (Hwang et al., 1998; Lu et al., 1998).
Histochemical observations (Figure 5) suggested that the IDS2 promoter fragments respond to Fe deficiency in spatially different patterns depending on the proximal sequences fused. The 5′-deletion fragments of the IDS2 promoter containing the native TATA box directed expression predominantly in Fe-deficient root endodermis, pericycle, and cortex, but not in Fe-sufficient roots, or in Fe-sufficient and Fe-deficient leaves in transgenic tobacco (Yoshihara et al., 2003). Although expression patterns conferred by the 3′-deletion or mutation sequences fused to 35SΔ46 (Figure 5a–c) were largely similar, expression in epidermis and root cap was observed only in the clones carrying the 35S minimal promoter in place of the native TATA box of the IDS2 promoter. Weak expression in Fe-deficient leaf veins and Fe-sufficient root cortex was also observed in clones carrying the distal region upstream of −273 of the IDS2 promoter. 3′-Deletion analysis suggested that minor cis-acting elements conferring Fe-deficiency-induced expression in leaves and constitutive expression in roots might exist in −491/−273 and −1696/−1541, respectively (Figure 1). A sequence within 35SΔ46 might synergistically support the function of these elements. Alternatively, a proximal sequence downstream of the TATA box in the IDS2 promoter might restrict the tissue specificity and suppress the expression under Fe sufficiency.
Furthermore, the clones carrying the IDS2 promoter sequences fused to 35SΔ90 expressed GUS strongly in an even greater range of tissues, including the epidermis of Fe-deficient roots (Figure 5d). These clones also expressed GUS in Fe-deficient leaves (Figure 2), and clones 11-90 and 14-90 also responded to Fe deficiency in leaves. These expression patterns were not seen with the IDS2 promoter fragments alone, nor with 35SΔ90 alone, suggesting a synergistic interaction. The −90/−47 region of the CaMV 35S promoter contains a cis-acting sequence called as-1, which drives expression preferentially in roots (Benfey et al., 1989). This sequence exhibits multiple synergistic interactions when combined with other elements, inducing an expression pattern that is different from that of the native individual elements. These synergistic interactions not only confer responsiveness to various stimuli, such as light (Benfey and Chua, 1989; Donald and Cashmore, 1990) and pathogens (Rushton and Somssich, 1998), but also drive expression in diverse organs, including germinating seeds (Benfey et al., 1990a; Salinas et al., 1992), stems and leaves (Benfey et al., 1990b; Lam and Chua, 1990; Salinas et al., 1992), and mature seeds (Yoshihara et al., 1996). Therefore, in the case of the IDS2 promoter fragments fused to 35SΔ90, IDE2, IDE1, and possibly other unidentified elements in the −272/−91 sequence might have generated complex synergism to the as-1 sequence.
Sequence comparison of IDE1, IDE2, and other Fe-deficiency-inducible promoters suggests conservation of cis-acting elements in various genes and species
We investigated whether sequences similar to IDE1 and IDE2 exist in other Fe-deficiency-inducible gene promoters (Figure 7). IDE1 has a 10-bp palindromic sequence (AAGCATGCTT; bold box in Figure 7), which is a candidate for the core sequence. We found that the HvNAAT-A promoter possesses a conserved sequence strikingly similar to IDE1 and its 3′-flanking 4 bp. This sequence was also conserved almost completely in the HvNAAT-B promoter. Therefore, the 3′-flanking sequence of IDE1 could make some minor contribution to its function, although 3′-deletion analysis demonstrated no pronounced effect of deletion of this region (Figure 2). Moderately conserved sequences to IDE1 were also found in the HvNAS1, HvIDS3, OsNAS1, and OsNAS2 promoters. Therefore, all the identified promoters of barley and rice genes that participate in the biosynthesis of MAs and respond to Fe deficiency share sequences similar to IDE1. The promoter of another Fe-deficiency-inducible gene OsIRT1, which encodes a putative ferrous Fe transporter (Bughio et al., 2002), also possesses sequences homologous to IDE1. Conserved cis-acting elements may be responsible for the co-ordinated induction of Fe-uptake-related genes. Homologous sequences found at −190/−207 and −309/−292 reside within the −348/−171 region of the HvNAS1 promoter that was suggested to contain putative Fe-deficiency-responsive elements (Higuchi et al., 2001b).
Sequences similar to IDE1 were also found in Fe-deficiency-inducible gene promoters of A. thaliana (Strategy I plant). Of these, the promoter of a ferrie-chelate reductase gene (AtFRO2) contained sequences notably similar to IDE1. A sequence homologous to the palindromic sequence was also found in the AtIRT1 promoter, although this was situated outside the 1-kb upstream region of the AtIRT1 promoter that was previously shown to be sufficient for Fe-deficiency-responsiveness in Arabidopsis (Vert et al., 2002). The observation that genes involved in both Strategy-I and Strategy-II Fe-acquisition mechanisms may share homologous cis-acting sequences, along with the finding that IDS2 and the HvNAS1 gene promoter of barley (Strategy II) can function in transgenic tobacco (Strategy I), suggests that partially conserved cis/trans systems are responsible for Fe-deficiency-inducible gene expression in plant species with different Fe-acquisition mechanisms, possibly involving some variation in the cis-sequences and their arrangement and the trans-acting factors. The promoter of AtNAS1, one of the Arabidopsis nicotianamine synthase genes (Suzuki et al., 1999), also contains a sequence homologous to IDE1, although no Fe-deficiency response of AtNAS1 expression has been reported to date.
Interestingly, the upstream region of the IDS2 promoter itself contained a region homologous to IDE1 at −772/−789. Although 5′- or 3′-deletion of −772/−789 and its neighboring sequences caused no substantial change in expression (Yoshihara et al., 2003; Figure 1), this sequence might be able to complement the function of IDE1. Furthermore, IDE2 was also homologous to IDE1 and its 3′-flanking 4 bp. Substantial homology between IDE2 and other Fe-deficiency-inducible gene promoters was found only in the sequence conserved in IDE1 and IDE2. This suggests that IDE1 and IDE2 share functional sequences. Duplication of either IDE1 alone or IDE2 alone, instead of the pair, IDE1 and IDE2, might confer responsiveness to Fe deficiency. Gain-of-function experiments will clarify whether IDE1, IDE2, and the upstream sequence at −772/−789 are functionally equivalent or not.
We also investigated whether IDE1 and IDE2 are similar to other known cis-acting elements. Previously identified cis-acting elements involved in Fe-deficiency-induced expression include the ferric iron uptake repressor (Fur) box of Gram-negative bacteria (de Lorenzo et al., 1987; Straus, 1994), the AFT1-binding site of yeast (Yamaguchi-Iwai et al., 1996), and the iron-responsive element (IRE) of vertebrates (Theil, 1990). In higher plants, transient assay analyses have identified two types of cis-acting elements that de-repress the expression of phytoferritin genes via Fe loading: the iron-dependent regulatory sequence (IDRS) in maize and the Arabidopsis ferritin gene promoters (Petit et al., 2001), and the iron regulatory element (FRE) in the soybean ferritin gene promoter (Wei and Theil, 2000). IDE1 and IDE2 did not show similarity to these elements. Recently, a gene encoding a basic helix-loop-helix (bHLH) transcriptional regulator (fer) was isolated using map-based cloning of the Fe-inefficient mutant T3238fer of tomato (Ling et al., 2002). IDE1 and IDE2 also do not share consensus binding sites with bHLH (CANNTG). Therefore, IDE1 and IDE2 are novel cis-acting elements that confer responsiveness to Fe deficiency. IDE1 and IDE2 are also the first cis-acting elements identified as being responsible for micronutrient-deficiency-inducible expression in higher plants. We found no known root-specific elements similar to either IDE1 or IDE2. Partial sequences of IDE1 and IDE2 showed some similarity to the copper-responsive trans-acting factor (ACE1)-binding site of the yeast metalothionein gene (TCY(4−6)GCTG; Thiele, 1992). A search of the PLACE database (Higo et al., 1999) revealed that IDE2 possesses two reported plant cis-acting sequences (shown with broken lines in Figure 7): the GTGA motif in the tobacco g10 promoter, which confers preferential expression in pollen (Rogers et al., 2001), and the binding site of a potato silencing binding factor (YTGTC(A/T)C) present in pathogenesis-related gene promoters (Boyle and Brisson, 2001). However, there is no evidence for the involvement of these elements in Fe-deficiency-inducible expression.
Understanding the mechanisms regulating Fe metabolism and creating plants tolerant to Fe deficiency
Understanding the mechanism regulating Fe metabolism in higher plants is still restricted at the physiological level. In Strategy-I plants, the Fe-deficiency responses are regulated by both shoot-borne signals and Fe availability at the rhizosphere or root itself (Schmidt, 2003). However, the actual signal substances are yet to be identified. Only recently, genes that may be involved in Fe-signaling pathways were isolated using map-based cloning of mutants defective in normal Fe-deficiency responses (Ling et al., 2002; Rogers and Guerinot, 2002). Using microarray techniques, some transcription factors were shown to be induced by Fe deficiency (Negishi et al., 2002; Wang et al., 2002). Nevertheless, the actual roles of those genes in the molecular regulation of the Fe-deficiency response are still unclear. The cis-acting elements first identified in this report, IDE1 and IDE2, will serve as powerful tools to clarify the molecular mechanisms that regulate Fe-deficiency-inducible gene expression and the underlying mechanisms regulating Fe metabolism. We are currently attempting to isolate trans-acting factors that interact with the defined cis-acting elements. Genetic engineering of stress-related transcription factors is a potent method for producing stress-tolerant plants, as demonstrated by Kasuga et al. (1998) who successfully improved drought, salt, and freezing tolerance of Arabidopsis. In contrast, our initial attempt to produce plants with enhanced tolerance to low Fe availability involved introduction of barley genomic fragments containing HvNAAT genes and their native promoters into rice (Takahashi et al., 2001). Manipulation of the cis-acting elements and trans-acting factors involved in Fe metabolism will become an attractive approach for producing crops that are even more tolerant to Fe deficiency and carry other favorable traits.
Construction of the chimeric genes and transformation of tobacco plants
The promoter-GUS regions of the chimeric constructs are shown schematically in Figures 1, 2, 3(b), and 4(b). The chimeric constructs introduced into tobacco were derived from either pLP19 (Szabados et al., 1990) or pBI101 (Clontech, Palo Alto, CA, USA), both of which contain the kanamycin-resistance gene cassette and the uidA (GUS) gene followed by the nopaline synthase terminator inside the T-DNA region. The −90/+8 (relative to the transcriptional start site) region of the CaMV 35S promoter (35SΔ90) is located upstream of the GUS gene of pLP19. The −46/+8 region of the CaMV 35S promoter (35SΔ46) was amplified by PCR using the following primers: 35S-46F: TCTAGAGGATCCAGCCCCGCAAGACCCTT and 35S-46R: CCCGGGTGTAATTGTAATTGTAAATAG.
The amplified fragment was cloned into pCR2.1TOPO (Invitrogen, Carlsbad, CA, USA), and the sequence was confirmed by DNA sequencing (ABI PRISM 310, PE Applied Biosystems, Tokyo, Japan). The verified fragment was excised using BamHI and SmaI sites designed at the 5′-ends of primers 35S-46F and 35S-46R, respectively, and inserted upstream of the GUS gene in pBI101. The resultant plasmid was named pBI101d46.
The promoter region of the IDS2 gene was isolated from barley (Hordeum vulgare L. cv. NK 1558), as described previously by Okumura et al. (1994). The 3′-deletion fragments (−1696/−91, −1696/−273, −1696/−492, −1696/−686, −1696/−866, −1696/−1129, −1696/−1274, −1696/−1541, −272/−91, −272/−136, −272/−181, and −272/−227 from the translation start site) were synthesized by PCR using the following primers:
• −1696F: AAGCTTGGATTGGGGATAAACACCTC
• −272F: AAGCTTCGAGGACGATTTGAACGGC
• −91R: GGATCCATCCTCCGGCCAGCCGTG
• −273R: GGATCCTCGTCTAGTATGGATTGTTA
• −492R: GGATCCTAGTCCAAGTTATTTATTCA
• −686R: GGATCCACTACTTCTATATGTATGCG
• −866R: GGATCCAGGGAATCACAAGCCTGT
• −1129R: GGATCCATGTTGAATGAGTATTTAGA
• −1274R: GGATCCTACATGAAAATTCAGATTGA
• −1541R: GGATCCAAGATGGTCTTAGCCTCC
• −136R: GGATCCGCAAGAAGCATGCTTGATGA
• −181R: GGATCCTGTCTGATCAAATCATGCGT
• −227R: GGATCCGAGAGTGGGAGTGACAGC
The amplified fragments were cloned into pCR2.1TOPO, and their sequences were confirmed by DNA sequencing. The verified fragments were excised using HindIII and BamHI sites at the 5′-ends of the forward (F) and reverse (R) primers, respectively, and inserted at the HindIII and BamHI sites of either pBI101d46 or pLP19.
The mutation fragments of the IDS2 promoter were created by annealing and filling-in pairs of synthetic oligomers that were complementary at the 3′-ends. Several fragments with or without a sequential 9-bp transversion (mutation from A to C, C to A, G to T, and T to G) were designed from the sequence of the −180/−136 region (Figure 3a). They contained BamHI sites at their proximal ends and BglII, XhoI, or HindIII sites at their distal ends. The fragments based on the −272/−181 or −272/−227 sequences contained HindIII sites at their distal ends. A sequence in either −185/−181 (AGACA) or −231/−227 (CTCTC) was changed to either AGATCT or CTCGAG using a 2-bp substitution and a 1-bp insertion to produce a BglII or XhoI site at the junction with the −180/−136 fragment, respectively (Figures 3a and 4a). Several mutation fragments with a sequential 9- to 10-bp transversion were also designed from the sequence of the −272/−227 region (Figure 4a). For the mutation at −235/−227, a HindIII site was introduced in place of the XhoI site at the junction with the −180/−136 fragment (Figure 4a, clone M25-46). Each pair of oligomers for the −180/−136, −272/−181, or −272/−227 sequences were annealed, filled in to give double-stranded DNA using the Klenow fragment, gel-purified, digested with the corresponding restriction enzyme, and cloned into pBluescript SK (Stratagene, La Jolla, USA) or pCR2.1TOPO. The fragments with the desired sequences were selected by DNA sequencing, and then inserted at the HindIII and BamHI sites of pBI101d46.
The constructed plasmids were introduced into Agrobacterium tumefaciens strain C58 by electroporation and used to transform tobacco (Nicotiana tabacum L. cv. Petit-Havana SR1) using the standard leaf-disk method (Helmer et al., 1984). Regenerated T0 transformants were transplanted in soil and grown in a greenhouse at 30 ± 3°C under a 12-h light/12-h dark cycle until the T1 seeds were harvested. The presence of the chimeric genes was confirmed by PCR amplification.
Growth conditions for the GUS assay
The T1 transformants were grown under conditions of Fe deficiency and the GUS activity was measured using the method by Yoshihara et al. (2003) with slight modifications. Seedlings of the T1 transformants were screened and grown on MS medium (Murashige and Skoog, 1962) containing 3% sucrose, 0.2% gellan gum, and 100 mg l−1 kanamycin for about 20 days at 28 ± 1°C with a 16-h light/8-h dark cycle. Then, five seedlings per line were transplanted onto rafts in a membrane culture system (Life Raft, pore size: 25 µm; Life Technologies, Rockville, MD, USA), which was floated on MS liquid media with or without Fe. Roots and leaves of two plantlets for each line were harvested 6, 7, 20, or 30 days after transplanting onto the membrane and used for the GUS assay.
Fluorometric and histochemical analyses of GUS activity
Assays of GUS activity were performed according to Jefferson (1987), as described earlier by Yoshihara et al. (2003). The enzyme activity was measured fluorometrically using 4-methylumbelliferyl-β-d-glucuronide as the substrate and the reaction product 4-methylumbelliferone (MU) was detected. The protein concentration was determined with a Bio-Rad Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA) and GUS-specific activity was expressed as pmol MU min−1 mg−1 protein. The significance of differences between each experimental block was analyzed using the non-parametric Mann–Whitney test (Ichihara, 1990).
For histochemical observation, the roots were dipped in GUS-staining solution (1 mm X-Gluc in 100 mm phosphate buffer (pH 7.0), 10 mm Na2-EDTA, 1 mm potassium ferricyanide, 1 mm potassium ferrocyanide, and 0.1% Triton X-100) and kept at 37°C overnight. Stained roots were excised and fixed in 4% paraformaldehyde, 5% glutaraldehyde, 0.1 m CaCl2, and 0.1 m cacodylate buffer (pH 7.0) for 3 h on ice. Fixed samples were serially dehydrated in ethanol and acetone, and embedded in Spurr's resin. Semi-thin sections (5 µm) were cut and slightly stained with toluidine blue and basic fuchsin (epoxy tissue stain; Electron Microscopy Sciences, Fort Washington, PA, USA) for light microscopy.