The increase in the ratio of root growth to shoot growth that occurs in response to phosphate (Pi) deprivation is paralleled by a decrease in cytokinin levels under the same conditions. However, the role of cytokinin in the rescue system for Pi starvation remains largely unknown. We have isolated a gene from Arabidopsis thaliana (AtIPS1) that is induced by Pi starvation, and studied the effect of cytokinin on its expression in response to Pi deprivation. AtIPS1 belongs to the TPSI1/Mt4 family, the members of which are specifically induced by Pi starvation, and the RNAs of which contain only short, non-conserved open reading frames. Pi deprivation induces AtIPS1 expression in all cells of wild-type plants, whereas in the pho1 mutant grown on Pi-rich soils, AtIPS1 expression in the root was delimited by the endodermis. This supports the view that pho1 is impaired in xylem loading of Pi, and that long-distance signals controlling the Pi starvation responses act via negative control. Exogenous cytokinins repress the expression of AtIPS1 and other Pi starvation-responsive genes in response to Pi deprivation. However, cytokinins did not repress the increase in root-hair number and length induced by Pi starvation, a response dependent on local Pi concentration rather than on whole-plant Pi status. Our results raise the possibility that cytokinins may be involved in the negative modulation of long-distance, systemically controlled Pi starvation responses, which are dependent on whole-plant Pi status.
Phosphorus is one of the most limiting nutrients for plants because Pi, the preferentially assimilable form, is unevenly distributed in soils, and >80% is immobile and not readily available for roots (Holford, 1997; Marschner, 1995). Plants have evolved adaptive responses to conditions of Pi limitation, responses that are operative at both metabolic and morphological levels (for review see Raghothama, 1999).
Many of the morphological responses to Pi starvation, such as an increase in root/shoot ratio and in lateral root number and length, are similar to those elicited by nitrogen starvation, and both are reminiscent of the effects of alterations in the levels of auxins or cytokinins. A decrease in the content of cytokinin, a hormone favouring shoot growth while inhibiting lateral root proliferation, has been associated with Pi starvation as well as with N deprivation (Horgan and Wareing, 1980; Salama and Wareing, 1979; Wagner and Beck, 1993). Moreover, exogenously supplied cytokinins counteract the root-growth stimulation induced under a low-nutrient regime (Kuiper, 1988; Kuiper et al., 1988). It is also known that Pi starvation rescues the root-hair defect of the auxin-resistant axr2 mutant (Bates and Lynch, 1996), while the auxin transport inhibitor CMPA prevents root-hair growth (Bates and Lynch, 1996). In addition, localized application of Pi or nitrate stimulates lateral root growth locally, and represses growth in the rest of the Pi or nitrate-starved roots (Drew, 1975; Robinson, 1994; Zhang and Forde, 1998), mimicking auxin and cytokinin effects, respectively. Lateral root proliferation induced by localized nitrate application is prevented in the axr4 mutant which is impaired in auxin response (Zhang et al., 1999). Finally, in N-starved plants the expression of several ‘two-component’ signalling genes can be rapidly induced by the supply of either cytokinin or nitrate (Sakakibara et al., 1998; Taniguchi et al., 1998).
How these nutrient starvation-induced effects on root architecture converge with other Pi starvation-specific responses is largely unknown. The identification of genes specifically induced by Pi starvation (reviewed by Raghothama, 1999) provides a starting point to address this question. It has already been shown that two types of genes that are induced by Pi starvation are inhibited throughout the whole root system following localized Pi fertilization, in a similar way to the repression of root development following localized Pi fertilization (Burleigh and Harrison, 1999; Liu et al., 1998). One of these types of systemically repressed genes are the members of the TPSI1/Mt4 family, a class of genes that have the characteristics of Pi starvation-signalling genes. They display highly specific responsiveness to Pi starvation, their RNAs share a conserved nucleotide motif, and they encode only short, non-conserved reading frames. These properties suggest that they are riboregulators or encode signal peptides (Burleigh and Harrison, 1997; Liu et al., 1997).
Here we report on the isolation and characterization of a novel member of the TPSI1/Mt4 family, AtIPS1 from Arabidopsis, whose RNA shows limited and non-conserved protein-coding potential. We demonstrate the specific responsiveness of AtIPS1 expression to Pi starvation, and provide evidence that the long-distance systemic signal controlling the Pi starvation response acts via negative control operating at overall high Pi status of the plant, rather than through an activation signal in response to low Pi. We also show that AtIPS1 and other Pi starvation-inducible genes are repressed by cytokinins in roots. However, cytokinins do not repress the Pi starvation-induced increase in root hair number and length, a response dependent on local Pi concentration. Our results point to a possible involvement of cytokinins in the negative modulation of Pi starvation responses that are dependent on whole-plant Pi status, and are systemically down-regulated.
Isolation of AtIPS1, a novel member of the TPSI1/Mt4 gene family from Arabidopsis
To identify key regulatory components involved in the Pi-starvation response, we isolated genes that respond to Pi-limiting conditions through differential screening. Using two cDNA probes, made from RNAs of plants grown first for 5 days in complete medium, then for 2 days under either Pi-starvation or Pi-sufficient regimes, we screened a cDNA library prepared from Pi-starved Arabidopsis plants (approximately 1′5 × 105 p.f.u.). Thirty-one cDNA clones were identified, displaying a strong differential hybridization signal, and were grouped into two classes (AtIPS1 and AtIPS2) based on high-stringency hybridization and restriction mapping. Sequencing and database searches revealed that AtIPS1 and AtIPS2 were related to each other and belonged to the TPSI1/Mt4 family (Burleigh and Harrison, 1999). AtIPS2 was identical to the previously described At4 gene, except for an 8 nt insertion in AtIPS2, GTGTGTGT (at nt 606 of At4, co-ordinates as given by Burleigh and Harrison, 1999). This difference is probably due to interecotypic polymorphisms. As in the TPSI1/Mt4 family (Burleigh and Harrison, 1997; Burleigh and Harrison, 1999; Liu et al., 1997), the AtIPS1 cDNA is predicted to contain only short (non-conserved) ORFs (Figure 1a). Among those starting with an AUG codon, the longest ORF is 24 amino acids long, and none of the ORFs was shared by other TPSI1/Mt4 genes, except a single ORF also found in At4, which is the most highly related gene to AtIPS1. This shared ORF encodes a four-amino-acid peptide, MAIP. In addition, AtIPS1 contains an 80-amino-acid ORF which does not start at an AUG codon, and which is not shared by any other gene of the TPSI1/Mt4 family. At the nucleotide level, AtIPS1 and At4 share two regions of sequence similarity of 251 and 96 nucleotides, respectively, localized at the 3′ and 5′ ends of the cDNA (Figure 1b). Overall nucleotide similarity in these regions is ≈70%, but in the 3′ region a stretch of 57 nucleotides is identical, except for a single base mismatch. Within this highly conserved nt stretch, a core of 22 nt is shared (95% identity) with all other known members of the TPSI1/Mt4 family: TPSI1, Mt4 and At4.
As shown in Figure 2, AtIPS1 RNA accumulated strongly in roots and shoots of plants subjected to Pi starvation (Figure 2a). Transcript accumulation was detected in both roots and shoots 2 days after Pi starvation, and continued to increase until after at least 7 days' stress. Similar results were obtained for At4. The expression of At4 observed in whole plants in response to Pi starvation was different from a previous report, which showed root-specific Pi starvation responsiveness of this gene (Burleigh and Harrison, 1999). As a control for our expression analysis we tested the expression of the Pi transporter gene AtPT1, and found it to be induced specifically in roots (Figure 2a), as described previously (Muchhal et al., 1996), indicating that under our conditions there is a proper organ regulation of the Pi starvation response. The TPSI1 gene from tomato has also been reported to be expressed both in roots and aerial parts of Pi-starved plants (Liu et al., 1997).
We also analysed AtIPS1 RNA in the pho1 mutant, which has impaired xylem loading of Pi and consequently has a reduced concentration of Pi in its leaves (Poirier et al., 1991). As expected, AtIPS1 was expressed in the leaves of the pho1 mutant (Figure 2b). The Pi control of this gene was examined further by evaluating the effect of resupplying Pi for different durations to the starved plants. After 8 days' growth under Pi-sufficient conditions, the previously Pi-starved plants showed a highly reduced level of AtIPS1 RNA (Figure 2c). We conclude that the induction of AtIPS1 expression under Pi starvation is reversible.
We have also studied the expression of AtIPS1 in response to other stresses such as K- and N-starvation, salinity and H2O2, and to the phytohormones abcisic acid, gibberellic acid, auxin and cytokinin, and during normal development in different organs. In no case was AtIPS1 RNA detected, as shown in Figure 2(d) for the nutritional starvation stresses, emphasizing the specificity of its response to Pi starvation. Similar results were observed for At4 under all conditions tested for AtIPS1.
An AtIPS1:GUS fusion in transgenic Arabidopsis is specifically responsive to Pi starvation
An AtIPS1:GUS translational fusion was prepared and transformed into Arabidopsis plants. We first screened an Arabidopsis genomic library using the AtIPS1 cDNA as a probe and isolated an AtIPS1 genomic clone. A 2.8 kb DNA fragment, containing the promoter and part of the AtIPS1 transcribed region (until nt 29 of the sequence shown in Figure 1a), was fused in-frame to the coding region of the GUS reporter gene present in the PBI101.2 plasmid (Jefferson et al., 1987). T2 progeny of 19 Arabidopsis transgenic T1 plants were grown in Pi-starvation or Pi-sufficient conditions and in situ GUS activity was determined. In 14 lines a clear increase of GUS activity in response to Pi starvation was observed (not shown). In none of the lines was GUS activity detected in response to any of the other stresses or phytohormones tested. Homozygous plants from one transgenic Arabidopsis line were used for in situ analysis of GUS activity during Pi starvation. Plants grown under Pi-sufficient conditions for 5 days and then starved for 1 day showed GUS staining in cotyledons and roots. In main roots GUS activity was confined to the region surrounding the elongation zone (Figure 3a). Plants that were Pi-starved for 7 days showed strong GUS staining in leaves and in all parts of the roots, including the region of the main root where lateral roots are formed.
AtIPS1:GUS responsiveness to Pi starvation in roots is repressed by cytokinins
To evaluate whether cytokinins or auxins had any effect on the expression of AtIPS1:GUS, transgenic plants were grown for 5 days on Pi-sufficient medium and then transferred to plates with or without Pi, and with or without cytokinins or auxins. As shown in Figure 3(b), cytokinin had a striking effect on AtIPS1:GUS response to Pi starvation. Indeed, this hormone repressed the GUS staining induced by Pi starvation in the root, whereas GUS staining in the aerial part was unaffected or slightly enhanced. This effect of cytokinin was also observed when the plants were grown in aerated liquid medium, and was independent of whether kinetin, zeatin or 6-benzylaminopurine (BA) was used (data not shown). On the other hand, given the well known effect of cytokinins on ethylene biosynthesis and the possible role of this latter hormone on the control of Pi-starvation responses (Lynch and Brown, 1997; Vogel et al., 1998), we evaluated whether ethylene would repress the AtIPS1:GUS gene, and found no evidence for this. Growing Pi-starved transgenic plants in the presence of the ethylene precursor ACC (1-amino cyclopropane-1-carboxylic acid; 1–30 µm), had no effect on AtIPS1:GUS expression (data not shown).
The effects of auxins were moderate and more complex, and their biological significance needs further examination. After 24 h starvation in the presence of the auxin indole-3-acetic acid (IAA), reporter plants showed GUS staining preferentially in the vascular tissue throughout the whole root, instead of being restricted to the region surrounding the elongation zone. However, in the presence of auxin GUS activity was not progressively extended to the rest of the main root after 7 days of Pi starvation, and the staining remained restricted to lateral roots and to the elongation zone of the main root.
Cytokinins also repress the response of other Pi starvation-inducible genes in roots
To confirm the results on the activity of the AtIPS1 promoter with the reporter gene, and to evaluate the effect of cytokinins and auxins on other Pi starvation-responsive genes, we performed Northern analysis. As expected, cytokinins repressed AtIPS1 RNA accumulation in the root (Figure 4a). Similar results were observed with all other three Pi starvation-responsive genes tested, At4, AtPT1 and AtACP5, which encodes an acid phosphatase (del Pozo et al., 1999). These results suggest that cytokinins may have a broad role in the regulation of the Pi-starvation response.
Auxins provoked only a moderate increase of AtIPS1 RNA accumulation at early stages of the Pi starvation stress, whereas at later stages the effect on RNA level was negligible in quantitative terms (Figure 4a), in line with results obtained with the reporter gene. Similar results were obtained for all the other Pi starvation-inducible genes examined, with the exception of AtPT1, expression of which was found to be reduced rather than augmented by auxins at early stages of the Pi starvation process.
Cytokinins do not repress the stimulation of root hair length by Pi starvation
Members of the Pi transporter and of the TPSI1/Mt4 family have been shown to be systemically down-regulated by localized application of Pi (Burleigh and Harrison, 1999; Liu et al., 1998). In contrast, the stimulation of root-hair elongation by Pi starvation is not responsive to this systemic signal, as it is dependent only on the local concentration of Pi (Bates and Lynch, 1996). In order to test whether there is any correlation between susceptibility to long-distance systemic down-regulation and to repression by cytokinin, we studied the effects of cytokinin on root-hair length. As shown in Figure 5, cytokinins did not repress this response, underlining the parallel between long-distance repression and the effects of cytokinins. The cytokinin treatment resulted in an increase of root hairs relative to the untreated controls, independent of the Pi growth regimen, which might be explained by invoking the known effect of cytokinins on ethylene biosynthesis, and of this latter hormone on root hair growth (Vogel et al., 1998)
Root expression of the AtIPS1:GUS reporter gene in the Pi xylem-loading mutant pho1 is delimited by the Casparian band
pho1 plants are affected not in Pi uptake from the soil, but in xylem loading (Poirier et al., 1991). Therefore it could be predicted that in this mutant the genetically induced Pi starvation status would not affect all cells: for instance, root epidermal cells would not be starved. To test whether the Pi-starvation response could be elicited in non-starved cells under conditions of overall low Pi status of the plant, the AtIPS1:GUS reporter gene was introduced into the pho1 mutant background through crossing and selfing. Contrasting with the observed lack of GUS activity in wild-type transgenic plants grown on Pi-rich soil, pho1 mutant plants harbouring the gene AtIPS1:GUS displayed GUS staining in roots and leaves when grown under these conditions. Remarkably, GUS activity in roots was restricted to the vascular bundle (Figure 6). Detailed analysis of root sections showed that GUS activity was confined to the endodermis, a tissue that is apoplastically isolated from the outermost cell layers by the Casparian band (Figure 6b). Thus GUS staining in the pho1 mutant is exclusively restricted to cells presumed to be suffering from Pi-starvation stress.
Using a newly identified member of the highly specific Pi starvation-responsive TPSI1/Mt4 gene family, AtIPS1, we have demonstrated an influence of cytokinins on the expression of AtIPS1. Cytokinin represses AtIPS1 transcript levels and also reduces the expression of several other Pi starvation-responsive genes, providing a first indication at the molecular level that cytokinins may play a broader role than previously suspected in the regulation of the Pi-starvation response.
AtIPS1, like other members of the TPSI1/MT4 family, is highly specifically induced by Pi starvation, has non-conserved, short ORFs, and shares a striking nucleotide sequence motif of 22 bp with other members of the family, characteristics suggestive more of a role as a riboregulator than as a structural gene. The 22 bp conserved motif is part of a common reading frame present in the tomato and Medicago truncatula members of the family, TPSI1 and Mt4, respectively (Burleigh and Harrison, 1997; Liu et al., 1997). Neither AtIPS1 nor At4 from Arabidopsis shares this ORF (or another ORF starting with ATG) within this region. This suggests that the 22 nt signature motif of members of this family may provide some important structural features, either for the biological activity of the members of this family (for example in their action as possible riboregulators), or for the control of their biological activity. However, At4 and AtIPS1 contain reading frames that span the conserved 22 nt motif, of 60 and 80 amino acids, respectively. In the case of AtIPS1, the 80 amino acid reading frame – unrelated to those of the other TPSI1/Mt4 members – does not start by AUG. There are precedents for eukaryotic RNAs that encode peptides in which the AUG codon does not serve as a translational start site (Yang et al., 1993). It remains to be demonstrated whether translation of this region is required for the function of these genes.
It is noticeable that At4 and AtIPS1 share the first ORF found in their cDNAs, which encodes a small peptide of four amino acids (MAIP). According to the Kozak model (Kozak, 1986) and to Lütke et al. (1987), the AUG start codon of this small ORF should be efficiently translated; our AtIPS1:GUS fusion was made including this AUG. However, this ORF is not present in the Medicago or tomato TPSI1/Mt4 RNAs, which share an ORF encompassing the conserved 22 nt motif (Burleigh and Harrison, 1997; Liu et al., 1997). The significance of the conservation of different ORFs between Mt4 and TPSI1 on the one hand, and At4 and AtIPS1 on the other, remains to be established. One possible explanation could be the existence of some degree of functional redundancy between the 3′ and 5′ parts of the RNAs of these genes, as has been suggested for members of the ENOD40 family, of another type of gene lacking large ORFs (Charon et al., 1997).
The specificity of AtIPS1 responsiveness to Pi starvation is indicated by the fact that, among those tested, no other stress or hormone induces its expression. The analysis of GUS activity of transgenic plants harbouring an AtIPS1:GUS reporter fusion showed that, at earlier stages of the starvation stress, the reporter gene was preferentially expressed in cotyledons, the first organ from which Pi mobilization is expected to occur, and in the area surrounding the elongation zone of the root. This region of the root is highly active in growth, a condition likely to accelerate Pi depletion. At prolonged periods of Pi starvation, GUS activity was observed throughout the whole plant. Thus it appears that induction of AtIPS1 can occur in any cell, provided it is Pi starved. The results obtained with the pho1 mutant plants harbouring the AtIPS1:GUS fusion support this interpretation. This mutant is impaired in Pi xylem loading (Poirier et al., 1991) and, in line with a strict requirement of Pi starvation at the cell level, in a Pi-rich soil, the activity of the GUS reporter gene in the roots of these mutant plants was restricted to the endodermis. Thus despite an overall low Pi content in this mutant, non-starved cells in the root (such as those of the epidermis and cortex) do not display GUS activity. This conclusion was not reached by Burleigh and Harrison (1999), probably because in their study At4 expression in the roots of the pho1 mutant was examined using RT–PCR of whole-root RNA, whereas in our study we were able to examine expression at the cell level. Due to the Casparian strip, Pi can be transported only throughout the endodermis towards the stele using the symplastic pathway (Marschner, 1995), suggesting that the PHO1 gene may be involved in this process.
Several of the morphological responses elicited by Pi starvation, and to a great extent shared with responses to N deficiency, are reminiscent of the action of the hormones auxin and cytokinin (Bates and Lynch, 1996; Marschner, 1995; Zhang et al., 1999). Here we have observed that specific responses, such as the induction of AtIPS1, At4, AtACP5 and AtPT1, are also influenced by these hormones. The effects of auxins are complex, and their biological significance remains to be shown. Thus the shift in the pattern of GUS staining towards the vascular tissue, provoked by auxins at early stages of Pi-starvation stress, might simply reflect an indirect effect; for example, auxins provoke a shift in the type of cells first suffering Pi-starvation stress by inducing phloem unloading. This interpretation is consistent with our results with AtPT1, the expression of which was found to be reduced rather than augmented by auxins at early stages of the Pi starvation process (Figure 5b), and Pi transporters are known to be preferentially expressed in epidermis and root hairs (Raghothama, 1999).
In the case of cytokinins, the observed effects of reducing expression of Pi starvation responsive genes are likely to be of biological relevance, because this hormone also reduces root growth and the levels of cytokinin decrease in response to phosphate starvation (Horgan and Wareing, 1980; Salama and Wareing, 1979; Wagner and Beck, 1993). The repressor effect of cytokinins does not appear to be a consequence of changes in root development caused by this hormone, as it occurs in all root cell types and in root tissue formed both before and after cytokinin treatment. Neither does this effect appear to be related to alterations in source–sink relationships caused by this hormone (Roitsch, 1999). Cytokinins would be expected to reduce translocation of mineral nutrients from the shoot, as has been shown in the case of nitrogen nutrients (Simpson et al., 1982); root starvation would thus be exacerbated. One possibility is that cytokinins serve to integrate general nutrient starvation and Pi starvation-specific responses. This would represent a novel role for cytokinins beyond their participation in the control of cell-cycle, organogenesis and senescence-related processes (Mok and Mok, 1994). It will be interesting to test whether specific responses to nitrogen starvation are similarly under the control of this hormone.
There are parallels between the repression of gene expression promoted by cytokinins and the phenomenon of systemic down-regulation of the Pi-starvation responses following localized application of Pi fertilizer (Burleigh and Harrison, 1999; Drew, 1975; Drew and Saker, 1984). Most nutrient-deficiency responses in the roots take into account not only the local nutrient status, but also whole-plant nutrient status, so that if one region of the root system receives enough nutrient for the whole plant, the corresponding nutrient-starvation response will be systemically down-regulated in the remaining part of the root system. The parallelism between this phenomenon and cytokinin effects is evident from the fact that systemic down-regulation has been shown to operate for members of the TPSI1/MT4 family, as well as that of the high-affinity Pi transporters (Burleigh and Harrison, 1999; Liu et al., 1998), and here we show that representatives of both families are down-regulated by cytokinins. In addition, the Pi starvation-induced stimulation of root-hair number and size, a response that is known to be exclusively dependent on local Pi status, and therefore not subjected to long-distance down-regulation (Bates and Lynch, 1996), is not repressed by cytokinins (Figure 6). In the case of Pi starvation, Pi per se does not appear to be the systemic repressor signal in plants, because repression of the response occurs well before the non-fertilized roots reach a high Pi level (Burleigh and Harrison, 1999). However, it remains to be shown whether cytokinins are involved in systemic down-regulation. Work is under way to test this possibility through the identification and characterization of mutants altered in the responsiveness of the AtIPS1:GUS reporter gene to cytokinins.
Plant material and growth conditions
Arabidopsis thaliana (L.), ecotypes Landsberg and Columbia (the latter for experiments involving transgenic plants), were used in this work. Plants were grown in soil or in controlled growth media based on the MS salts, including those for nutrient-starvation stresses, as described previously (del Pozo et al., 1999), except for the experiment examining the effect of hormones on increases in root-hair length and number under Pi starvation. In this case the MS medium was replaced by Johnson's medium because in the latter the increase in root hairs, induced by Pi starvation, is much more evident (Bates and Lynch, 1996). The hormonal treatments were performed with the cytokinins kinetin, 6-benzylaminopurine (BA) or zeatin (1–10 µm); the auxin indole-3-acetic acid (IAA, 1–10 µm); or the ethylene precursor 1-amino cyclopropane-1-carboxylic acid (ACC, 1–30 µm), and were given to 5-day-old seedlings for 1 or 7 days in the presence or absence of Pi.
Standard molecular procedures
All methods were performed as described previously (Avila et al., 1993; Sambrook et al., 1989), except where indicated. The vectors for cloning were pBluescriptII (Alting-Mees and Short, 1989). To screen for genes induced under Pi, two cDNA probes were prepared from RNA extracted from plants grown for 7 days in complete medium or Pi-starved for the last 2 days. Probes were obtained from RNA reverse-transcribed in the presence of 32P ATP.
Preparation of an AtIPS1:GUS fusion and transformation of Arabidopsis plants
An AtIPS1:GUS fusion was made by cloning in-frame a 2.8 kb HaeIII fragment of the AtIPS1 genomic clone, containing the promoter region and part of the transcribed region (until nt 29 of the sequence shown in Figure 1a), with the GUS-coding region present in the binary vector pBI101.2 (Jefferson et al., 1987). Agrobacterium tumefaciens harbouring this construct was used to transform Arabidopsis ecotype Col-0 by in planta vacuum infiltration (Bechtold et al., 1993). Kanamycin-resistant T1 plants were selected by plating seeds on MS medium supplemented with 2% glucose and 50 mg ml−1 kanamycin, and transferred to soil.
Histochemical analysis of GUS activity
GUS activity was detected in situ in transgenic Arabidopsis plants, harbouring the AtIPS1:GUS fusion as previously described (Liang et al., 1989), except that root cross-sections performed on the pho1 mutant were embedded in 4% agarose for hand sectioning.
Computer programs for protein and nucleic acid analysis
We are grateful to Professors F. Salamini and C. Martin for critical reading of the manuscript and helpful suggestions. We also thank Dr Y. Poirier for providing us with the pho1 mutant, and M.J. Benito for her excellent technical assistance. J.C.d.P. and R.S. were recipients of a PhD fellowship from the Universidad Complutense and of a postdoctoral contract with the Ministry of Education, respectively. This research was financed by the EU (contract BIO4-CT96-0770).