Two genes with a common region that is characteristic of the TPSI1/Mt4 family were cloned from a Pi-starvation-induced cDNA library of rice roots using suppression subtracted hybridization (SSH). Based on the consensus sequence of these two genes, members of the TPSI1/Mt4 family were found in maize, wheat and barley. blast and a cluster analysis in the eight members of the TPSI1/Mt4 family showed two classes of four genes each among monocots. The first gene from rice was designated OsIPS1 based on a comparison of the consensus sequence with AtIPS1, and consequently the second gene, which has been previously reported as OsPI1, was designated OsIPS2. Accumulation of the mRNA of OsIPS1/2 was examined by northern blotting and quantitative reverse transcriptase-polymerase chain reaction in whole-root and split-root experiments under treatment with phosphate (Pi) and the Pi analogue phosphite (Phi). OsIPS1 showed much higher mRNA accumulation in roots than OsIPS2, and an opposite trend was seen in shoots. OsIPS1/2 showed both systemic and local responses to Pi starvation, and less than 10% of the overall induced mRNA level was due to the local Pi concentration in roots. The results indicate that Phi may interfere with earlier events in roots that are associated with a local Pi signalling pathway. An analysis of transgenic plants showed that OsIPS1/2 are independently responsive to Pi signalling and are mainly expressed in lateral roots and in the vascular cylinder in the primary root. Exogenous cytokinin (6-BA) almost completely suppressed systemic Pi starvation signalling and partially suppressed local Pi signalling. Exogenous abscisic acid remarkably reduced Pi starvation signalling. In contrast, exogenous auxin enhanced Pi signalling, especially local Pi signalling in roots. Exogenous ethylene (ethyphon) and the ratio of auxin to cytokinins did not appear to affect the expression of these two genes.
Although phosphorus (P) plays a crucial role in plant growth and development, it is also the least available plant macronutrient in many natural and agricultural ecosystems, especially in acidic and calcareous soils. To cope with this limited availability of P in the growth medium, plants have evolved sophisticated strategies that involve root morphological and metabolic changes to enhance phosphate (Pi) acquisition and utilization (Bates & Lynch 1996; Raghothama 1999). To design crops that handle Pi more efficiently, it is important that we understand the mechanisms that underlie the Pi-sensing and -signalling pathways involved in molecular and developmental responses to Pi starvation.
Over the past decade, several genes that are specifically responsive to Pi starvation have been found, including genes with unknown function in the TPSI1/Mt4 family, such as TPSI1, Mt4, At4, AtIPS1 and OsPI1 (Burleigh & Harrison 1997; Liu, Muchhal & Raghothama 1997; Martín et al. 2000; Wasaki et al. 2003). The genes in the TPSI1/Mt4 family are dramatically and specifically induced by Pi starvation, which makes them useful as marker genes for investigating Pi sensing and signalling in plants (Rubio et al. 2001). Split-root experiments and the pho1 mutant of Arabidopsis, which is unable to load Pi into the xylem have been used to investigate the responses of Mt4 (Medicago truncatula) and At4 (Arabidopsis). The results suggested that the suppression of these genes in roots was systemic through shoots; namely that long-distance signals controlled the response to Pi starvation in roots (Liu et al. 1998; Burleigh & Harrison 1999). By using AtIPS1, Martin et al. (2000) reported that exogenous cytokinins suppress the expression of AtIPS1 and other genes that respond to Pi starvation, but do not suppress the increase in the number and length of root hairs induced by Pi starvation, which is dependent on the local Pi concentration in roots rather than on the whole-plant Pi status (Bates & Lynch 1996). These results suggest that the response of genes to Pi starvation in roots may depend on both the whole-plant Pi status and the local Pi concentration in roots.
Using an Arabidopsis transgenic line with AtIPS1::GUS, a phr1 mutant, which is defective in Pi starvation-responsive gene expression, was identified, and the PHR1 gene has been shown to be a MYB transcription factor that is homologous to a P regulatory gene (PSR1) in Chlamydomonas and also to regulate AtIPS1 expression (Rubio et al. 2001). PHR1-binding sequences (GNATATNC, P1BS) are present in the promoter region of Pi-starvation-responsive structural genes, indicating that this protein acts downstream in the Pi starvation signalling pathway. As with other members of the TPSI1/Mt4 family, the function of AtIPS1 is unknown. Since aspects of the response to Pi starvation, such as altered root-hair development and changes in the root-to-shoot ratio, were not observed in phr1, the system that regulates the PHR1 gene, which is related to the physiological and morphological adaptation to Pi starvation in plants, is essentially unknown.
Recent studies in yeast and higher plants have suggested that phosphite (Phi; H2PO3– or HPO32–), a reduced analogue of Pi (H2PO4– or HPO42–), can suppress typical molecular and developmental responses to Pi starvation, although it cannot eliminate the nutritional need for Pi in plants (Plaxton & Carswell 1999; Ticconi, Delatorre & Abel 2001). It has been reported that Phi targets PHO84, a high-affinity Pi transporter and putative component of a Pi sensor complex in yeast (McDonald, Niere & Plaxton 2001), but does not affect the expression of auxin-inducible genes (Ticconi et al. 2001), which suggests that the suppression of Pi starvation responses by Phi is selective and not caused by a general cellular toxicity of this Pi analog. Thus, Phi may be a useful molecular tool for elucidating the details of Pi sensing in plants.
Phytohormones are important signals in root initiation and elongation. Auxin is believed to mediate root-hair elongation in response to Pi starvation (Bates & Lynch 1996), but is not involved in the enhancement of root-hair initiation by Pi starvation (Ma et al. 2001). It has been suggested that auxin signalling is unlikely to play a role in changes in root system architecture response to Pi starvation (Williamson et al. 2001), whereas roots grown under low-Pi conditions are more sensitive to auxin with respect to both the inhibition of primary root elongation and the promotion of lateral root development, than roots grown under high-Pi conditions (López-Bucio et al. 2002); that is, auxin can amplify Pi starvation signalling (Al-Ghazi et al. 2003). It has been reported that a decrease in ethylene production and a greater sensitivity of roots to ethylene occurred during aerenchyma formation under a temporary deprivation of phosphorus, which suggests that ethylene may play a role in plant responses to Pi starvation (Drew, He & Morgan 1989; He, Morgan & Drew 1992), but we do not yet know how this ethylene signalling pathway interacts with the Pi signalling pathway at the genetic and molecular level. It has been demonstrated that cytokinins are involved in the negative regulation of long-distance, systemically controlled responses to Pi starvation, which are dependent on the whole-plant Pi status (Salama & Wareing 1979; Martín et al. 2000), but it is still unclear whether cytokinins affect the responses to the local Pi concentration. Our present understanding of the interaction between the Pi signalling pathway and the hormone-signalling pathway that is involved in mediating the molecular and developmental responses to Pi starvation is incomplete.
To investigate the Pi signalling pathways that depend on the whole-plant Pi status and local Pi availability, and their interaction with hormones in rice, a genomic model dicotyledon, two genes of the TPSI1/Mt4 family were cloned from a Pi-starvation-induced cDNA library of rice roots using suppression subtracted hybridization (SSH) (Xia et al. 2002). The homologous genes in maize, wheat and barley were also identified for a cluster analysis together with AtIPS1, At4, Mt4 and TPSI1. The results indicate that there are two groups of four genes each among monocots and one group of four genes among dicots. According to a phytogenetic tree, the two genes in rice were designated OsIPS1 and OsIPS2, the latter of which has been previously reported as OsPI1 (Wasaki et al. 2003). Split-root experiments with hormone and phosphite treatment were performed to investigate the expression patterns of these two genes under different whole-plant Pi conditions and local Pi availability in roots, and to examine the effects of different hormones on their expression. Transgenic plants were developed and analysed to determine the P element in the promoter region of OsIPS1, and the possible functions of OsIPS1 and OsIPS2.
MATERIALS AND METHODS
Plant growth conditions
A japonica variety of rice (Oryza sativa L. cv. Nipponbare) was used and the plants were grown hydroponically in this study. For Pi starvation stress, rice seeds were germinated at 37 °C for 1 d and pre-cultured at 30 °C for 2 d. After pre-culture, the plants were grown in normal solution (320 µm Pi) (Yoshida et al. 1976) for 12 d. The 14-day-old seedlings were then transferred to nutrient solution without Pi (NaH2PO4 was replaced by NaCl)for 5 d and Pi was added to the solution. The pH of the culture solution was adjusted to 5.0 using 1 m NaOH every day. The plants were grown in a growth chamber under a photosynthetic photon flux density of approximately 200 µmol photons m−2 s−1 with a 16 h light (28 °C)/8 h dark (22 °C) photoperiod. Humidity was controlled at approximately 60%.
For other stresses, 14-day-old seedlings were grown in nutrient solutions with no nitrogen for 5 d, or under dehydration for 12 h, cold at 4 °C for 24 h, or salt stress with 180 m m NaCl for 24 h. Hormonal treatment was carried out by adding exogenous cytokinin [1 µm 6-benzylaminopurine (6-BA)], auxin (1 µm NAA), the auxin transport inhibitor NPA (5 µm), ethylene (the ethylene-producing compound ethephon, 10 µm), or 10 µm abscisic acid (ABA) to the culture solution with and without Pi for 14-day-old seedlings grown for 5 d in both whole-root and split-root experiments.
To investigate systemic and local Pi signalling, three split-root experiments were carried out: (1) half of the roots received 320 µm Pi (SR1/+Pi) and the other half were subjected to Pi starvation (SR1/–Pi); (2) half of the roots received 320 µm (Phi) (the Pi analogue phosphite) (SR2/+Phi) and the other half were subjected to Pi starvation (SR2/–Pi); and (3) half of the roots received 320 µm Pi (SR3/+Pi) and the other half received the Pi analogue phosphite (SR3/+Phi).
Seeds of maize, barley and wheat were germinated on filter paper at 30 °C for 3 d, and then grown hydroponically in Hoagland solution (Arnon & Hoagland 1940) for 12 d at pH 5.7. The plants were grown with or without Pi for 10 d and the roots and shoots were harvested separately on the fourth and tenth days for RNA isolation.
Cloning and characterization of OsIPS1 and identification of ESTs of related genes in monocots
Two positive EST clones (OsIPS1/2) were obtained from a Pi-starvation-induced rice root cDNA library constructed using SSH (Xia et al. 2002), and sequenced (MagaBASE 1000; Amersham Pharmacia, CA, USA). A sequence comparison revealed that the second clone was completely identical to the reported gene OsPI1 (Wasaki et al. 2003), and the full-length cDNA of OsIPS1 was then cloned using Pi-starvation-induced cDNAs as a template by 5′- and 3′-RACE (rapid amplification of cDNA ends) using a Marathon cDNA amplification kit (Clontech Laboratories, Palo Alto, CA, USA). The primer 5′-GGCAGGGCACACTC CACATATC-3′ was used for amplification for 5′-RACE and the primer 5′-CTCCCCAAGAAGAAGAAGTAGC TTAG-3′ was used for amplification for 3′-RACE. All of the gene clones and constructions were sequenced and homology searches were performed with the GenBank/EMBL database using the blast program. A cluster analysis and a phylogenetic tree analysis were performed using the Clustalw program (DNASTAR Inc., WI, USA)
DNA and RNA gel blot analysis
Genomic DNA was isolated from young leaves of rice using the cetyl trimethyl ammonium bromide (CTAB) method (Murray & Thompson 1980) and total RNA was extracted from the roots and leaves of plants using the Trizol D0410 reagent according to the procedure recommended by the manufacturer (Invitrogen, Carlsbad, CA, USA). Five micrograms of genomic DNA were digested with restriction enzymes DraI, EcoRI, EcoRV, HindIII and XbaI, and separated on 0.8% agarose gel. Twenty micrograms of total RNA were separated on 1.0% agarose gel denatured with formaldehyde. After electrophoresis, the digested DNA and total RNA were transferred to a Hybond-N+ nylon membrane (Amersham Pharmacia). 32P-dCTP-labelled cDNA was used as a probe. The blots were hybridized and washed at 65 °C under stringent conditions.
Construction of GUS fusion, over-expression and RNA-interference vectors and the development of transgenic plants
The OsIPS1 (2.1kb) and OsIPS2 (2.2kb) promoter sequences were polymerase chain reaction (PCR)-amplified using the primers 5′-GAGGAGAGAAGTTTTTCGGT GGAG-3′ and 5′-TGGGTGCTTTTATTTGGAAGTA TTG-3′, and 5′-AGGGGATGCAAAAGATGGTGTC-3′ and 5′-GTAGAGAGGAATCCAGAAGAAGGGTC-3′, respectively, with the genomic DNA of Nipponbare as a template. After being blunted, the promoter fragments were cloned in the sense orientation in the SmaI site of the vector pCAMBIA-1391Z. The OsIPS1 full-length sequence (5′-TACGAGCTCCACTTTTAACTAAAGAAATGATT TCCTG-3′ and 5′-TACGGATCCATACCAAGAAGTA ACTAGAGATACCC-3′) was PCR-amplified with the genomic DNA of Nipponbare as a template and inserted between SacI and BamHI of the vector pCAMBIA-1301 which was inserted at a CaMV(Cauliflower mosaic virus) 35S promoter (between EcoRI and BamHI) and Nos Poly A (between PstI and HindIII) for over-expression. The double-stranded RNA interference (RNAi) construct of OsIPS1 was ligated with forward and reverse fragments, into which had been inserted the second intron of NIR1 in maize. The gene fragment was amplified using the primers 5′-GGCAGGGCACACTCCACATTATC-3′ and 5′-CTCC CCAAGAAGAAGAAGTAGCTTAG-3′ with the genome DNA of Nipponbare as a template, and the forward fragment, intron and reverse fragment of the DNA were then cloned seriatim into the vector pCAMBIA-1301, into which was inserted a CaMV(Cauliflower mosaic virus) 35S promoter (between EcoRI and BamHI) and Nos Poly A (between PstI and HindIII). Agrobacterium tumefaciens (EHA105) harbouring these constructs was used to transform the rice cultivar Nipponbare as described by Chen et al. (2003).
Histochemical analysis and GUS assay of transient expression
Histochemical GUS analysis was performed according to Jefferson, Kavanagh & Bevan (1987). Transgenic plant root samples were incubated with X-gluc overnight at 37 °C. After being stained, the tissues were rinsed and fixed in FAA [formalin : acetic acid : 70% ethanol (1 : 1.2 : 17.8 in volume)] for 24 h, embedded in Spurr resin, and then sectioned. Sections (5 µm) were mounted on slides and photographed.
Quantitative real-time PCR
cDNA was synthesized from total RNA extracted from shoots and roots using a first-strand cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA). The cDNA samples were stored at −20 °C. Quantitative real-time PCR was performed with an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) using the SYBR green I master mix (Applied Biosystems) containing optimized buffer, dNTP, and Taq DNA polymerase. Each 20-µL reaction contained 2 × SYBR green master mix, 2 µL of cDNA, and 0.1 µL forward and reverse primers (20 µm). The primers used for quantitative PCR were 5′-AAG GGCAGGGCACACTCCACATTATC-3′ and 5′-ATTAG AGCAAGGACCGAAACACAAAC-3′ (for the OsIPS1 gene), 5′-CCT TCTTCTGGATTCCTCTC-3′ and 5′-AGTT CACCACAAAAGATACAGTAG-3′ (for the OsIPS2 gene) and 5′-GGAACTGGTATGGTCAAGGC-3′ and 5′-AGTCTCATGGATACCCGCAG-3′ (for the rice Actin gene). Reaction conditions for thermal cycling were: 95 °C for 2 min and 40 cycles of 95 °C for 15 s, 56 °C for 15 s, and 72 °C for 45 s. Fluorescence data were collected during the 72 °C step. Using this method, PCR product is monitored by measuring the increase in fluorescence caused by the binding of SYBR green dye to dsDNA. For each gene, a standard curve was established using a serial dilution of an experimental cDNA sample. Data were analysed with Sequence Detector version 1.7 software (Applied Biosystems). The software calculated the threshold cycle from the plot of the increase in intensity of fluorescence of the reporter dye versus the cycle number. The quantity of cDNA was calculated from the threshold cycle by interpolation from the standard curve. To account for differences in total RNA present in each sample, the amount of cDNA calculated was normalized using the amount of Actin cDNA detected in the same sample.
Measurement of the total Pi concentration in plant tissues
Shoots and roots were separately sampled from five plants at defined time points and under each treatment to measure the total Pi concentration. The samples were oven-dried at 80 °C for 3 d and ground. For the determination of the total Pi concentration, about 20 mg of each sample was digested with 1 mL of concentrated H2SO4 overnight and H2O2 was added at 120 °C for 30 min until the aliquots of the digested solution just became colourless. The resulting solutions were diluted with water to 50 mL, adjusted to pH 4.0–5.0 by 6 m NaOH, and the Pi contents were analysed by the phosphomolybdenum blue reaction using a Spectroquant NOVA60 spectrophotometer and Spectro-quant Phosphat-Test Kit (Merck, Darmstadt, Germany). The total Pi concentrations (Unit is mg Pi g−1 DW) in shoots and roots of each plant were calculated based on the dry weight of biomass (DW), which was taken for mensuration and the Pi content.
Cloning and characterization of the OsIPS1/2 genes
Two cDNA clones (P3A10 and P2D6) were identified from a Pi-starvation-induced cDNA library of rice roots that had been previously constructed in our laboratory using SSH. Northern blotting analysis was performed for the two clones at different time-points under Pi starvation in shoots and roots, and under different stresses, including Pi starvation, N starvation, dehydration stress, cold stress and salt stress. The results indicated that both clones were remarkably induced in roots after 2 d of Pi starvation stress, and this induction continued to increase until the end of treatment at 5 d (Fig. 1a). The mRNA of the second clone in shoots was detectable after 4 d of stress. Neither transcript was detected in + P shoots nor roots, and when P was re-supplied to –P plants, the induced expression rapidly disappeared (Fig. 1a). As shown in Fig. 1b, among the stresses tested, the two genes were specifically induced by Pi starvation.
Sequencing of the two clones and a blast search in the NCBI database (http://www.ncbi.nlm.nih.gov/BLAST) found characteristics common to the TPSI1/Mt4 family (AtIPS1, At4, Mt4 and TPSI1) (Fig. 2a) in the two clones. The second clone is identical to the gene OsPI1 reported by Wasaki et al. (2003). Full-length cDNA cloning was performed for the first clone (P3A10) using the 3′- and 5′-RACE methods, and the first clone was shown to include 661 bp with nine short open reading frames. In a comparison of the genomic sequences, no intron was found, similar to the results in OsPI1 (Wasaki et al. 2003). Southern blot analysis and in silico mapping indicated that the gene is a single-copy gene in rice, like OsPI1, and is located on chromosome 3 (Fig. 1c).
Comparison of the TPSI1/Mt4 family in monocots and dicots
By using primers based on ESTs that were found according to consensus sequences of OsIPS1 and OsPI1, two members of the TPSI1/Mt4 family in maize, wheat and barley were cloned. Like OsPI1, the seven genes cloned in this case were only found among monocots (URL: http://www.ncbi.nim.gov/dbEST/index.html). A cluster analysis showed that the eight genes in monocots could be grouped into two classes and the four genes in dicots could be grouped into one class (Fig. 2b). In the common characteristic regions of these genes, two boxes with eight bases each are conserved among all 12 of the TPSI1/Mt4 family members (Fig. 2a). The evolutionary relationship between the two groups in monocots is closer than that between monocots and dicots (Fig. 2b). The two classes of TPSI1/Mt4 family members in monocots are called IPS1 and IPS2, respectively. Based on the homology of the common region with AtIPS1, OsIPS1 named above belongs to the first group (OsIPS1(AY568759), ZmIPS1 (BM379264, CD058856), TaIPS1(BQ744479, BJ264395) and HvIPS1(AV911716)), and OsPI1 reported by Wasaki et al. (2003) belongs to the second group (OsIPS2(BI796836), ZmIPS2 (AY105557), TaIPS2(A709853) and HvIPS2(BJ447206, BJ453908)) and thus OsPI1 is referred to as OsIPS2 hereafter. Northern blotting analysis showed that HvIPS1, like OsIPS1, was more strongly induced by Pi starvation than HvIPS2 in roots. In contrast, TaIPS2 and ZmIPS2 were more strongly induced than TaIPS1 and ZmIPS1 (Fig. 2c).
Quantitative real-time PCR analysis of the expression of OsIPS1/2
To quantitatively investigate the temporal expression of OsIPS1/2, quantitative real-time PCR analysis was used for shoots and roots of 14-day-old seedlings subjected to Pi starvation at different time points. The expression patterns of the two genes based on quantitative reverse transcriptase (RT)-PCR were consistent with those based on Northern blotting. The induction of OsIPS1 was detectable after 4 h of starvation in roots and after 12 h in shoots. After 5 d of starvation, the amount of mRNA of OsIPS1 was increased by 2.5% in shoots, and by 100% in roots. When Pi was re-supplied to –P plants at 5 d, only about 1 and 2% of induced mRNA was found in shoots and roots at 6 d, respectively (Fig. 3b). OsIPS2 mRNAwas detected after 12 h in roots and after 24 h in shoots. The levels of OsIPS2 mRNA were only about 25% of those of OsIPS1 in roots at 24 h and beyond (Fig. 3b). In contrast, the levels of OsIPS2 mRNA in shoots were two to three times higher than those of OsIPS1 at the time points shown, and about 50% of those in roots at 1 d and beyond (Fig. 3b). Wasaki et al. (2003) reported that twice as much mRNA of OsPI1(OsIPS2) was detected in –P roots than in shoots at 72 h and beyond, which is consistent with the present results. Our results demonstrate that both OsIPS1 and OsIPS2 are mainly induced in roots and the expression level of OsIPS1 is much higher than that of OsIPS2 in roots, whereas an opposite trend is seen in shoots.
The concentration of total Pi in both shoots and roots was measured at all of the time points. A significant (P < 0.01) decrease in the total Pi concentration was found after 4 h of starvation in roots, but not in shoots (Fig. 3a). At 12 h and beyond, the Pi concentration with –P treatment in both shoots and roots continuously decreased by about 40% of that in the control at 5 d. The concentration with +P treatment in shoots slightly increased, but almost no change was seen in roots during the experiment. When Pi was re-supplied to the plants at 5d, the total Pi concentration measured at 6 d was increased by about 80% of that with +P treatment in both shoots and roots (Fig. 3a). The results suggest that the local Pi concentration in roots results in the earlier response of OsIPS1 to Pi signalling at 4 h, since no change in the Pi concentration occurred in shoots at 4 h. The strong induction of the two genes mainly depended on a starvation signal based on the whole-plant Pi status whereas a local Pi stress signal still exists. To confirm this assumption, 14-day-old whole seedlings and shoot-excised seedlings were grown under Pi starvation for 4 h for quantitative RT-PCR analysis. The results revealed that the local Pi concentration in roots promoted induction of OsIPS1 gene at 4 h (Fig. 3c).
OsIPS1/2 showed a similar temporal expression pattern, but a different spatial pattern. To determine whether these two genes respond independently to Pi starvation signals, the expression pattern of OsIPS2 was investigated in transgenic plants that showed the over-expression and RNAi of OsIPS1. The results indicate that the expression pattern of OsIPS2 was independent of that of OsIPS1 (Fig. 3d).
Tissue-specific expression patterns of OsIPS1/2 in roots
To investigate the tissue-specific expression patterns in roots, promoters of OsIPS1/2 were isolated and fused with the reporter gene GUS to develop transgenic plants. GUS staining indicated that both OsIPS1/2 were strongly expressed in lateral roots and their expression was also detected in the vascular cylinder of roots. GUS staining was detected in leaves driven by the promoter of OsIPS2, but not by promoter of OsIPS1, which is consistent with the results of northern blotting (Fig. 4a & b). GUS staining in cross-sections of primary root and lateral root, and in the transverse sections of an emerged lateral root indicated that OsIPS1 was expressed in phloem and pericycle cells in the vascular cylinder of primary root (Fig. 4c), and was expressed in all cells except for epidermal cells in lateral root (Fig. 4d). A transverse section of an emerged lateral root primordium showed that lateral root primordium emerged from the pericycle cell opposite the primary phloem with GUS staining.
Effects of systemic and local Pi signalling on the expression of OsIPS1/2
To further investigate the effects of systemic and local Pi signalling on the expression of OsIPS1/2, whole-root and split-root experiments were performed using 14-day-old seedlings subjected to treatments for 5 d. Three split-root experiments were designed: (1) half of the roots received Pi (SR1/+Pi) while the other half were subjected to Pi starvation (SR1/–Pi); (2) half of the roots received Phi (SR2/+Phi) while the other half were subjected to Pi starvation (SR2/–Pi); and (3) half of the roots received Pi (SR3/+Pi) while the other half were subjected to Phi (SR3/+Phi). The total Pi concentration in shoots and roots was measured for the experiments (Table 1). Northern blotting and quantitative RT-PCR analysis were performed to investigate the expression patterns of OsIPS1/2 under different conditions.
Table 1. The total Pi concentration (mg Pi g−1 DW) of rice grown in a root-split experiment with Pi or Phi treatment (n = 5)
SR1, split-root experiment 1 with P-supplied side (SR1/+P) and P-deficient side (SR1/–P); SR2, split-root experiment 2 with Phi-supplied side (SR2/+Phi) and P-deficient side (SR2/–P); SR3, split-root experiment 3 with P-supplied side (SR3/+P) and Phi-supplied side (SR3/Phi).
8.76 ± 0.29
8.12 ± 0.22
3.61 ± 0.24
3.15 ± 0.11
6.09 ± 0.53
7.44 ± 0.27
3.09 ± 1.10
4.33 ± 0.25
5.58 ± 0.59
3.84 ± 0.15
4.66 ± 0.61
3.14 ± 0.63
6.31 ± 0.45
7.08 ± 0.17
5.10 ± 0.73
The total Pi concentration in roots that were subjected to Pi starvation in the split-root experiment (SR1/–P) was similar to that in the whole-root experiment under Pi starvation, whereas the mRNA levels of OsIPS1/2 were decreased by more than 90% in roots that were subjected in Pi starvation in the split-root experiment (Table 1 and Fig. 5a & b). The mRNA was scarcely detected in the roots that received Pi, and the total Pi concentration was twice that in the roots subjected to Pi starvation and similar to that in the whole-root experiment under sufficient Pi. These results strongly suggest that the level of mRNA induced by local Pi signalling in roots is less than 10% of that caused by systemic signalling through the whole plant.
The Pi analogue phosphite (Phi), which is not involved in P metabolic pathways in plants, was used in whole-root and split-root experiments. In a whole-root experiment with Phi, the total Pi concentration in shoots and roots was about 50 and 70%, respectively, of the values in the whole-root experiment with sufficient Pi, indicating that less Phi is absorbed and translocated than Pi. The Pi concentration in shoots from the Phi-supplied experiment was only 50% of that in the sufficient-Pi experiment, but the level of mRNA induced in the former was only about 10% of that under Pi starvation (Table 1 and Fig. 5b). The total Pi concentrations in shoot and roots under Pi starvation in the split-root experiment (SR2) were similar to those in the whole-root experiment under Pi starvation (Table 1), whereas the level of mRNA induced in the former case was about 25% of that in the latter case (Fig. 5b), which indicates that the effect of Phi on Pi signalling is not dependent on systemic Pi signalling which involves the whole-plant Pi status.
The notion that less than 10% of the induced mRNA level in roots is mainly controlled by the local Pi concentration in roots under Pi starvation is also supported by the results of the SR3 experiment, in which half of the roots were treated with a high concentration of Pi whereas the other half were treated with Phi. The total Pi concentration in the shoots and roots that received Pi were similar, as in the SR1 and SR3 experiments, but the concentration in the roots that received Phi was higher than that in roots that underwent Pi starvation (Table 1). Consequently, the mRNA level in the roots that received Phi in the SR3 experiment was much lower than that in the roots that underwent Pi starvation in the SR1 experiment. These results indicate that the Pi concentration in the growth medium can affect Pi signalling in roots regardless of the whole-plant status (Fig. 5b).
Regulation of OsIPS1/2 expression by hormones
By northern blotting analysis, exogenous ethylene (ethephon), auxin (NAA) and the auxin transport inhibitor NPA did not affect the Pi-starvation-induced expression of OsIPS1/2, whereas cytokinin (6-BA) suppressed such expression, which is consistent with a report on AtIPS1 (Martin et al. 2000). Exogenous ABA remarkably reduced the mRNA levels of the two genes under Pi starvation (Fig. 6a). To quantitatively evaluate the effects of these hormones on Pi signalling that depends on the systemic Pi status as well as on local Pi signalling in roots, whole-root and split-root experiments were performed with the four hormones using quantitative RT-PCR analysis (Fig. 6b & c). The results indicate that the general changes in the mRNA level in the split-root experiment were consistent with those in the whole-root experiment, but the percentage change was different with the different hormones. The addition of 1 µm NAA to the roots that were subjected to Pi starvation resulted in a doubling of the mRNA levels of the two genes, whereas only about a 20% increase was found in the whole-root experiment (Fig. 6b & c). Exogenous cytokinin (1 µm 6-BA) severely reduced the mRNA level in the roots that underwent Pi starvation by about 80% compared with that without 6-BA in the split-root experiment, although this mRNA was scarcely detected in the whole-root case (Fig. 6b & c). These results suggest that cytokinin may almost completely suppress systemic Pi signalling that depends on the shoot Pi status and may partially suppress local Pi signalling in roots. Exogenous ethylene (10 µm ethephon) did not affect the responses of the two genes to Pi signalling. Northern blotting indicated that the ratio of NAA to cytokinins does not influence the expression of OsIPS1/2 and that the suppression of Pi starvation signalling by cytokinins is independent of auxin (Fig. 6a).
Among dicots, homologues from closely related species share greater amino acid identity than homologues from distantly related species (Burleigh & Harrison 1999). However, each species shows one member in each class in monocots (Fig. 2b), indicating that monocots may contain two members of theTPSI1/M4 family. The novel member in rice cloned in this study was grouped into the class that is more homologous with conserved sequences of AtIPS1, and therefore this gene was designated OsIPS1. Accordingly, the gene that was identical to the reported gene OsPI1 (Wasaki et al. 2003) was designated OsIPS2. There are only two conserved boxes of eight bases each among all 12 genes in dicots and monocots, and OsIPS1/2 share greater base identity with the two members in other monocots (Fig. 2a). Recently, two more members of the TPSI1/M4 family in Arabidopsis were identified as At4-1 and At4-2 (NCBI: AY536062, AY334555). It is possible that these new members may also exist in other dicots.
To detect possible subtle differences in responses to Pi starvation signalling between OsIPS1/2, quantitative RT-PCR was used to determine the relative mRNA levels of the two genes at different Pi starvation time points in leaves and roots. Except for the common characteristic expression that is specifically and drastically induced by Pi starvation and the stronger induction in lateral roots (Figs 1a, 4a & b), different expression patterns were found between OsIPS1/2: (1) OsIPS1 showed a much higher expression level than OsIPS2 in roots, and (2) although the induced expression of OsIPS2 in roots was lower than that of OsIPS1, the induced expression of OsIPS2 in shoots was much higher than that of OsIPS1. At 4 h of Pi starvation, induced mRNA of OsIPS1, but not OsIPS2, could be detected by quantitative RT-PCR (Fig. 2b). Wasaki et al. (2003) reported that induced mRNA of OsPI1 (OsIPS2) could be detected at 4 h of Pi starvation. These different results may be caused by different levels of Pi in the culture solution for the growth of rice seedlings. Whereas Wasaki et al. grew seedlings with 1 mg P L−1 before Pi starvation, we used a concentration of 320 µm Pi (10 mg P L−1) in our study. The delayed accumulation of a detectable level of OsIPS2 mRNA in our study may be due to a shielding effect of the initially sufficient whole-plant Pi status. The expression pattern of OsIPS2 in transgenic plants that over- and under-expressed OsIPS1 based on RNAi of OsIPS1 indicated that the two genes respond independently to the Pi starvation signal (Fig. 3d). The expression patterns of other members of the TPSI1/Mt4 family in monocots were also investigated. The induced expressions of OsIPS1, HvIPS1, ZmIPS2 and TalIPS2 were stronger than those of OsIPS2, HvIPS2 ZmIPS1 and TalIPS1 in roots after 4 or 10 d of Pi starvation, whereas higher induced expressions of OsIPS2, HvIPS1, ZmIPS1 and TalIPS1 were detected in shoots. These results regarding the temporal and spatial expression patterns of these group members are not consistent with phylogenetic classes based on the common regions of the TPSI1/Mt4 family. In fact, two copies of a putative cis-acting element that responds to Pi signalling, the PHR1 binding site (Rubio et al. 2001), were found in promoter regions of OsIPS1/2, but the locations of the element sites are different in the two promoters (at −567 and −432 in OsIPS1, and at −192 and −122 in OsIPS2). It is unclear whether the different binding site locations or the spatial structure near the sites causes the difference in the response to Pi signalling. In addition to the difference in the PHR-1 binding site, four more cis-acting elements were found in the promoter of OsIPS2 (Wasaki et al. 2003), but only one of them (NIT2) was also found in OsIPS1. It is possible that the other cis-acting elements play a role in the differences between the expression patterns of the two genes. The results with cross-sections of primary and lateral roots of transgenic plants with promoters of OsIPS1/2 and GUS fusion support the hypothesis that the induced expression of OsIPS1/2 may be involved in the initiation and development of lateral roots in rice (Fig. 4). However, at least in the seedling stage, RNAi and over-expression of OsIPS1 did not cause a detectable change in root growth under both Pi sufficiency and Pi starvation. The function of this specific response to Pi starvation and cell-specific expression of the gene require further elucidation.
It has been suggested that the whole-plant Pi status and the local Pi concentration in roots play important roles in the responses to Pi starvation in roots (Martin et al. 2000). Our results in whole-root and split-root experiments with Pi and phosphite (Phi) revealed that less than 10% of the overall accumulation of mRNA after 5 d of starvation can be attributed to local Pi signalling, and that early mRNA accumulation within 4 h is caused by a decrease in the root total Pi concentration (Fig. 3 BC), which may be important as a physiological and developmental adaptation to Pi starvation, such as root-hair initiation and elongation. It has been demonstrated that systemic Pi signalling that depends on the whole-plant Pi status and local Pi signalling induced by the Pi status in roots coexist, and this is consistent with our results. Our data also show that the earlier response is caused by local Pi signalling, and this response accounts for less than 10% of the overall induced expression over 5 d under Pi starvation.
Phosphite has been suggested to be a selective inhibitor of Pi starvation responses because although it cannot substitute for Pi to satisfy the nutritional need for phosphorus in plants, it suppresses typical molecular and developmental responses to Pi starvation (Ticconi et al. 2001). Plaxton & Carswell (1999) have proposed that phosphite interferes with early events in Pi sensing and signalling that initiate and co-ordinate cellular responses to Pi starvation. Our results showed that the induced mRNA level in roots of plants supplied with Phi was only about 10% of that in plants under Pi starvation (Fig. 5b). The total Pi concentrations in shoots and roots that received Pi in the SR3 experiment (while the other roots received Phi) were similar to those in SR1 (while the other half of the roots underwent Pi starvation), but the induced mRNA level in roots that received Phi was barely detected compared to that in roots that underwent Pi starvation (Table 1 and Fig. 5b), which strongly supports the hypothesis that Phi interferes with local events in Pi signalling and consequently blocks systemic Pi signalling in shoots.
Martin et al. (2000) found that exogenous cytokinins suppress the expression of AtIPS1 in response to Pi starvation, but do not suppress the increase in root-hair number and length induced by Pi starvation which is dependent on the local Pi concentration. Therefore, they suggested that cytokinins are involved in the negative modulation of the systemically controlled response to Pi starvation that depends on the whole-plant Pi status. However, our results indicated that exogenous cytokinin (1 µm 6-BA) may completely suppress Pi signalling that depends on the whole-plant Pi status and partially suppress local Pi signalling in roots in rice (Fig. 6). A cytokinin signal transduction pathway has been shown to consist of a two-component circuit in Arabidopsis (Hwang & Sheen 2001). Franco-Zorrilla et al. 2002) found that at 1 µm, a cytokinin (kinetin) suppressed AtIPS1::GUS nearly completely, but had no effect on cre1 mutants. However, a higher concentration of kinetin (5 µm), reduced AtIPS1::GUS even in cre1 mutants. Their results suggest that cytokinin may affect Pi signalling later in the two-component circuit in the cytokinin signal transduction pathway, and although CRE1 may be the main receptor for a low concentration of cytokinin (1 µm), other CRE genes may be involved in the translocation of cytokinin for higher concentrations of cytokinin (5 µm). In rice there are five putative cytokinin receptors (tblastn[http://www.ncbi.nlm.nih.gov/BLAST]), and RNAi of these receptor genes is carried out for the further investigation of the inter-crossing between P signalling and cytokinin signalling.
In contrast to cytokinin, the mRNA level in root that was subjected to 1 µm exogenous NAA and Pi starvation was about 20% higher than that in root that was subjected to only Pi starvation. A split-root experiment showed that the mRNA level that depended on local Pi signalling in roots was more than doubled after 1 µm exogenous NAA was added to roots that were subjected to Pi starvation (Fig. 6). This result suggests that while auxin is positively involved in the Pi starvation signalling that depends on the systemic Pi concentration, local Pi signalling in roots may be much more sensitive to regulation by auxin. Pi starvation stress increases root-hair density and length (Bates & Lynch 1996; Borch et al. 1999; Ma et al. 2001), which is believed to be controlled by a local Pi signal (Martin et al. 2000) and by auxin and ethylene (Schiefelbein 2000). The negative regulation by cytokinin and positive regulation by auxin of Pi signalling at both the systemic and local levels suggest that NAA signalling may crossover with local Pi signalling in roots that controls the root-hair elongation induced under Pi starvation. The auxin-induced expression of OsIPS1/2 is mediated by a cytokinin-independent pathway, and vice versa. It has been suggested that ethylene may be involved in the effects of Pi starvation on root-hair density and length (Schiefelbein 2000), while exogenous ethylene did not affect the local Pi concentration signal in roots in this study (Fig. 6).
ABA has been suggested to mediate the inhibitory effects of a high NO3– concentration on lateral root formation that depends on systemic negative feedback, and may also play a negative role in mediating the auxin-independent stimulation of lateral root elongation by the local application of NO3– at the post-emergence stage (Signora et al. 2001; Smet et al. 2003). Since auxin has been shown to be involved at this point, it is suspected that ABA may suppress the response of lateral root primordia to auxin. In this case, ABA had a negative regulatory effect, mainly on systemic Pi signalling that depended on the whole-plant Pi status (Fig. 6). In Arabidopsis, it has been demonstrated that exogenous ABA inhibits lateral root development and this ABA-induced lateral root inhibition is mediated by an auxin-independent pathway (Smet et al. 2003). Further studies are needed to answer the question of whether ABA signalling crosses over with Pi signalling in the regulation of the lateral root architecture in rice.
This research was supported by the National Key Basic Research Special Foundation of China (G1999011700), National Education Ministry of China and Natural Science Foundation of China.