Genetic and Genomic Evidence That Sucrose Is a Global Regulator of Plant Responses to Phosphate Starvation in Arabidopsis

Plants respond to phosphate (Pi) starvation by exhibiting a suite of developmental, biochemical, and physiological changes to cope with this nutritional stress. To understand the molecular mechanism underlying these responses, we isolated an Arabidopsis ( Arabidopsis thaliana ) mutant, hypersensitive to phosphate starvation1 ( hps1 ), which has enhanced sensitivity in almost all aspects of plant responses to Pi starvation. Molecular and genetic analyses indicated that the mutant phenotype is caused by overexpression of the SUCROSE TRANSPORTER2 ( SUC2 ) gene. As a consequence, hps1 has a high level of sucrose (Suc) in both its shoot and root tissues. Overexpression of SUC2 or its closely related family members SUC1 and SUC5 in wild-type plants recapitulates the phenotype of hps1 . In contrast, the disruption of SUC2 functions greatly inhibits plant responses to Pi starvation. Microarray analysis further indicated that 73% of the genes that are induced by Pi starvation in wild-type plants can be induced by elevated levels of Suc in hps1 mutants, even when they are grown under Pi-sufﬁcient conditions. These genes include several important Pi signaling components and those that are directly involved in Pi transport, mobilization, and distribution between shoot and root. Interestingly, Suc and low-Pi signals appear to interact with each other both synergistically and antagonistically in regulating gene expression. Our genetic and genomic studies provide compelling evidence that Suc is a global regulator of plant responses to Pi starvation. This ﬁnding will help to further elucidate the signaling mechanism that controls plant responses to this particular nutritional stress.

Although plants require ample amounts of phosphate (Pi) for their growth and development, they often encounter a Pi deficiency in their surrounding environments (Raghothama, 1999;Vance et al., 2003). Unlike animals, plants are sessile organisms, so they must respond to this adverse condition through adjustment in their growth and development and in metabolic activities. Through the long evolution process, plants have developed sophisticated strategies to better adapt to Pi starvation (Yuan and Liu, 2008).
These strategies include changes in the root architecture system (i.e. reduction of primary root growth and the formation of more lateral roots and root hairs), increased expression of Pi transporter genes, the induction and secretion of organic acid, RNase, and acid phosphatases (APases), and the accumulation of starch and anthocyanin. Although these adaptive responses have been widely studied and well documented in a variety of plant species, the molecular mechanisms that regulate these responses are still largely unknown.
In the last decade, significant progress has been made in identifying signaling components that are involved in plant responses to Pi starvation (Yuan and Liu, 2008). PHOSPHATE STARVATION RESPONSE1 (PHR1) is one of the most studied transcription factors; it encodes a MYB domain-containing protein that binds to a short DNA motif commonly found in the promoters of several Pi starvation-induced (PSI) genes (Rubio et al., 2001). The mutation of PHR1 causes defects in a subset of Pi starvation responses. Other transcription factors that are involved in Pi responses include WRKY75, WRKY6, ZAT6, MYB26, and BHLH32 in Arabidopsis (Arabidopsis thaliana) and OsPTF1 and OsPHR2 in rice (Oryza sativa; Chen et al., 2009;Lin et al., 2009). In Arabidopsis, the siz1 mutant exhibits exaggerated Pi starvation responses (Miura et al., 2005). SIZ1 is a SUMO E3 ligase that is localized in the nucleus. In vitro sumoylation assay has shown that PHR1 is the direct target of SIZ1. Besides several PSI genes, PHR1 also controls the transcription of miRNA399. miRNA399 can directly bind to the 5# untranslated region of PHO2 mRNA and result in its degradation (Bari et al., 2006). The PHO2 gene encodes a ubiquitin E2 conjugation enzyme and is involved in the regulation of plant Pi homeostasis (Bari et al., 2006;Chiou et al., 2006). Interestingly, the activity of miRNA399 can further be controlled through a mechanism called "target mimicry" (Franco-Zorrilla et al., 2007). In Arabidopsis, AtIPS1 and At4 genes encode non-protein-coding transcripts that function in the internal translocation of Pi from shoot to root (Shin et al., 2006). Both of them contain a motif that is partially complementary to miRNA399. The imperfect pairing of AtIPS1/At4 with miRNA399 results in the sequestration of miRNA399, thus altering the mRNA level of the PHO2 gene. Recently, some SPX domain-containing proteins have also been shown to act as important regulators of Pi starvation responses in both Arabidopsis and rice Liu et al., 2010a).
Sugar sensing and signaling are important regulatory components of plant growth and development as well as metabolic activities (Rolland et al., 2006). Experimental evidence has suggested that sugar signaling may also be involved in plant responses to Pi starvation. Stem girdling of white lupin (Lupinus albus) to block the movement of photosynthates from shoot to root severely affected the induction of PSI genes in Pi-starved root (Liu et al., 2005). Similarly, it has been observed that the level of induction of PSI genes in Arabidopsis seedlings is positively correlated with the concentration of Suc present in the culture medium . Both groups have also noticed that light has a significant effect on the induction of PSI genes (Liu et al., 2005;Karthikeyan et al., 2007). When plants are grown in the dark without addition of Suc in the medium, the expression of PSI genes is significantly reduced; however, if exogenous Suc is supplied in the dark, the expression of PSI genes can be sustained to a certain degree. The similar effects of photosynthates on Pi starvation-induced cluster root formation in white lupin and the expression of miRNA399 in common bean (Phaseolus vulgaris) have also been observed (Zhou et al., 2008;Liu et al., 2010b). In contrast, the Arabidopsis pho3 mutant carries a defective copy of the SUCROSE TRANSPORTER2 (SUC2) gene, leading to substantially reduced transport of Suc from shoot to root and the suppression of induction and secretion of APase on its root surface (Zakhleniuk et al., 2001;Lloyd and Zakhleniuk, 2004). In addition, earlier studies have revealed that the expression of many genes involved in the synthesis, translocation, and degradation of Suc was altered during Pi starvation (Hammond et al., 2003;Vance et al., 2003;Wu et al., 2003;Misson et al., 2005;Mü ller et al., 2007). Furthermore, an increase in Suc biosynthesis in Pi-starved leaves has been observed in a variety of plants, including Arabidopsis, bean, barley (Hordeum vulgare), spinach (Spinacia oleracea), and soybean (Glycine max; Hammond and White, 2008). An increase in Suc concentration is thought to be an early response of plants to Pi starvation (Hammond and White, 2008). All these results suggested that Suc may play a signaling role during the plant response to Pi starvation. However, how broadly Suc can affect plant responses to Pi starvation has not been examined, and the question of how Suc acts at the molecular level to trigger multiple Pi responses remained to be answered.
In this work, we characterize an Arabidopsis mutant, hypersensitive to phosphate starvation1 (hps1), which showed enhanced sensitivity in almost all the tested parameters of plant responses to Pi starvation. Our genetic and molecular studies of this mutant, combined with genomic analysis, provide compelling evidence that Suc is a global regulator of plant responses to Pi starvation.

Identification of the hps1 Mutant
The hps1 mutant was identified from an Arabidopsis T-DNA activation library. This library was generated by transforming a transgenic line bearing an AtPT2-LUC construct (herein referred as the wild-type plant) with a pSuperTag2 activation vector (Koiwa et al., 2006;Supplemental Fig. S1). In this construct, a firefly luciferase (LUC) gene is fused to the promoter of the AtPT2 (AtPht1;4) gene, which encodes a high-affinity Pi transporter (Karthikeyan et al., 2002). Pi deficiency induced the expression of the reporter gene fused to the Pi starvation-responsive AtPT2 promoter. When grown under Pi starvation, the hps1 mutant displayed enhanced LUC expression compared with wild-type plants (Fig. 1A). The enhanced expression of AtPT2-LUC in hps1 is rather restricted to the junction between root and hypocotyl and the root tip. To confirm that the enhanced expression of AtPT2-LUC was due to the mutation that occurred in a regulatory component, an AtPT2-GUS marker gene was introduced into hps1 plants through a genetic cross. Consistent with the results of AtPT2-LUC, expression of the AtPT2-GUS reporter gene was also greatly enhanced at the hypocotyl-root junction and root tip of hps1 (Fig. 1, B and C). Further examination indicated that enhanced AtPT2-GUS expression mainly occurred in lateral roots, while expression in the primary root was relatively suppressed compared with the wild type under the no-phosphorus (P2) condition. In hps1 mutants, the expression of AtPT2-LUC and AtPT2-GUS was evident even under Pi sufficiency (Fig. 1). This suggested a hyperinduction of AtPT2-LUC in the hsp1 mutant. To further ensure that the enhanced expression of AtPT2-LUC and AtPT2-GUS in hps1 mutants is a true reflection of the changes in expression of the endogenous AtPT2 gene, a real-time PCR technique was used to measure the mRNA level of AtPT2 in wild-type and hps1 plants. When the AtPT2 gene expression level of wild-type plants grown under the Pi-sufficient condition was set as 1, we found that induction of AtPT2 by Pi starvation in wild-type and hsp1 mutant plants was 4.2-and 30-fold, respectively (Fig. 2). Even though hsp1 mutant plants already had a high basal expression of AtPT2 (2.5 compared with 1 of the wild type), it still showed a higher induction (12.5fold) by Pi starvation than the wild type (4.2-fold).
When hps1 was back-crossed to a wild-type plant, the F1 progeny seedlings behaved like hps1. And the F2 progeny showed a 1:3 segregation of wild-type to mutant phenotypes (51 wild type versus 142 hps1). The pSuperTag2 plasmid harbors a Basta-resistant gene and a superpromoter that is located next to the T-DNA right border (Supplemental Fig. S1). In the F2 population, all the wild-type seedlings were Basta sensitive and mutants were resistant. These results indicated that the hps1 mutant phenotype is caused by a single dominant mutation, probably due to the T-DNA insertion.

The hps1 Mutation Enhances Pi Starvation-Induced Gene Expression
AtPT1 is another major high-affinity Pi transporter in Arabidopsis whose expression is also highly induced by Pi starvation (Muchhal et al., 1996;Karthikeyan et al., 2002). When hps1 mutation was introduced into the transgenic line carrying an AtPT1-GUS construct through a genetic cross, the expression of AtPT1-GUS was also highly induced under Pi starvation in the lateral roots around the hypocotyl-root junction and the root tips (Supplemental Fig. S2). The enhanced induction of the endogenous AtPT1 gene was confirmed by real-time PCR analysis (Fig. 2).
In addition to AtPT1 and AtPT2, we examined the expression of six other PSI genes that encode another Pi transporter (PHT1;5), an APase (ACP5), an RNase (RNS1), two noncoding mRNAs of At4 and AtIPS1, and miRNA399D. Similar to AtPT1 and AtPT2, the induction of these six PSI genes by Pi starvation was also dramatically enhanced in hps1 plants compared with wild-type plants (Fig. 2). In fact, even under Pisufficient conditions, the expression of these PSI genes (except AtPT1 and miRNA399D) in hps1 plants was higher than in wild-type plants. This indicated that the hps1 mutation alone is sufficient to induce some PSI genes even without Pi starvation. To respond to Pi starvation, plants increase the production and secretion of APases to mobilize internal and external Pi. The secreted APases on the root surface can cleave 5-bromo-4-chloro-3-indolyl phosphate (BCIP) to produce a blue precipitate (Zakhleniuk et al., 2001). When BCIP was applied to the root surfaces of Pi-starved wild-type and hps1 seedlings, hps1 roots showed much darker blue staining than the wild type (Supplemental Fig. S3A). In gel-assay of total extracted soluble proteins (for a detailed procedure, see "Materials and Methods") showed an enhanced production of a major isoform of APase (approximately 95 kD) in hps1 on both P+ and P2 media (Supplemental Fig. S3B). Consistent with the results of BCIP staining and in-gel assay, the total APase activity is enhanced in the hps1 mutant compared with the Figure 1. AtPT2-LUC and AtPT2-GUS expression in wild-type and hps1 seedlings. A, Seeds of wild-type (WT) and hps1 plants were directly sown on MS medium with (P+) or without (P2) supplemented inorganic Pi and grown for 8 d. Luciferin was sprayed on the surface of 8-d-old seedlings, and photographs were taken either with a digital camera (top row) or a CCD camera (bottom row). B and C, Expression of AtPT2-GUS in 8-d-old wild-type and hps1 seedlings. The GUS activities at the hypocotyl-root junction (B) and the root tip (C) are shown.
wild type, as shown by quantitative analysis using a spectrophotometer (Supplemental Fig. S3C).
Other characteristic responses of plants to Pi starvation include anthocyanin and starch accumulation. When hps1 and wild-type plants were grown on P2 medium for 2 weeks, the wild-type seedlings turned only light purple, whereas the whole shoot of the hps1 mutant had already become dark purple (Fig. 3A). Quantitative analysis showed that, under Pi starvation, the anthocyanin content in the hps1 seedlings was three times as much as in wild-type plants (Fig. 3B). In addition, hps1 plants have reduced amounts of chlorophyll a and b and carotene under both Pi-sufficient and Pi-deficient conditions compared with wild-type plants (Fig. 3C), suggesting a reduced photosynthesis activity. Our following microarray analysis indicated that six genes involved in photosynthesis were downregulated in hps1 mutants (Supplemental Table S1). The accumulation of starch in hps1 plants was also much higher than in the wild type under both P+ and P2 conditions, as revealed by iodine staining (Fig. 3D). All of the above results indicated that the hps1 mutation affects multiple aspects of plant responses to Pi starvation in addition to the expression of PSI genes.

Ion Homeostasis Is Perturbed in the hps1 Mutant
To sustain normal growth and development, it is important for plants to have a mechanism to maintain ion homeostasis, including Pi, as they respond to external stress (Raghothama, 1999). To examine the effects of hps1 mutation on ion homeostasis in plant cells, we analyzed the accumulation of nine elements in wild-type and hps1 plants under both Pi-sufficient and Pi-deficient conditions. When grown under Pi starvation, the contents of cellular Pi and total phosphorus in wild-type plants decreased dramatically in both their shoot and root tissues (Fig. 4, A and B). Similarly, when hps1 plants were grown under Pi starvation, they accumulated less cellular Pi and total phosphorus than on Pi-sufficient medium. However, hps1 plants contained more cellular Pi in roots and less in shoots under both P+ and P2 conditions compared with wild-type plants (Fig. 4A), indicating that the hps1 mutation caused a shift of allocation of Pi between shoots and roots. It has been reported that plants grown on Pi-deficient medium would accumulate more iron (Hirsch et al., 2006). Our analysis confirmed these observations and further showed that, on P2 medium, hps1 plants accumulated even more iron (about 2-fold higher compared with wild-type plants) in both shoots and roots (Fig. 4C). Another interesting phenomenon is that wild-type plants accumulated relatively high amounts of zinc ions in their roots compared with shoots on P+ medium (Fig. 4D). However, differential accumulation of zinc in shoots was not obvious in hps1 plants.
Besides phosphorus, iron, and zinc, we also examined the contents of sodium, potassium, calcium, magnesium, manganese, and sulfur in both wild-type and hps1 mutant plants under P+ and P2 conditions. We found that the concentrations of these ions were all more or less perturbed in hps1 plants (Supplemental Fig. S4).

The hps1 Mutation Affects Plant Root Development
In addition to plant responses to Pi starvation and ion homeostasis, we also found that the hps1 mutation had a marked effect on plant root development. When Figure 2. Quantitative analysis of PSI gene expression in wild-type and hps1 seedlings. Nine-day-old wild-type and hps1 seedlings grown on MS P+ or P2 medium were used for real-time PCR analysis. The PSI genes examined in this analysis encode three high-affinity Pi transporters (AtPT1, AtPT2, and PHT1;5), an APase (ACP5), an RNase (RNS1), two noncoding mRNAs (At4 and AtIPS1), and a microRNA (miRNA399D). Values are means 6 SE of three biological replicates where the fold changes are normalized to transcript levels in the wild type on P+ medium. Asterisks indicate significant differences compared with the wild type (P , 0.05, t test). White bars, Wild-type plants; black bars, hps1 plants.
grown on P+ medium, the roots of wild-type plants exhibited indeterminate growth during 15 d after seed germination. At day 9, wild-type plants seldom formed lateral roots (Fig. 5A). During the same growth period, the length of primary roots of hps1 plants was about half that of wild-type plants. hps1 plants also formed numerous lateral roots near the junction between hypocotyl and root (Supplemental Fig. S5). On P2 medium at day 9, the growth of primary roots of wild-type plants was strongly inhibited while several lateral roots were formed, as reported previously by another group (Williamson et al., 2001). In hps1 plants grown under P2 conditions, overall growth, including both shoots and roots, was severely inhibited. The length of the lateral roots was much shorter compared with that of the wild type, although there was not much difference in the number of lateral roots (Fig. 5, A and B). The hps1 mutation also had a strong effect on root hair development. As anticipated, wild-type plants grown on P+ medium developed few short root hairs close to the root tip and maturation zone (Fig. 5,B and C). When grown on P2 medium, both the number and length of root hairs increased in the wild type. The hps1 plants formed many short root hairs near the root tip and maturation zone, even under Pi-sufficient conditions (Fig. 5, B and C). When grown under Pi starvation, the increase in root hair length and density was less than that of the wild type.
Taken together, these data suggested that hps1 plants display a rather constitutive Pi starvation-induced developmental response under Pi sufficiency and become more sensitive to Pi starvation-induced inhibition of plant growth and development. When hps1 mutant plants were grown in soil, however, there was no obvious morphological difference compared with wild-type plants.
The hps1 Mutant Phenotype Is Caused by Overexpression of the SUC2 Gene Since the hps1 mutant phenotype is tightly linked to T-DNA insertion, we cloned the T-DNA-flanking sequence in hps1 plants. The T-DNA is inserted 170 bp upstream of the ATG start codon of the SUC2 gene ( Fig. 6A). Because of the presence of a superpromoter sequence next to the T-DNA right border (Koiwa et al., 2006), there was a likelihood of constitutive overexpression of the SUC2 gene in hps1 plants. Indeed, the mRNA level of SUC2 in hps1 was about 15-fold higher than that of wild-type plants, as revealed by real-time PCR analysis (Fig. 6B). [ 14 C]Suc uptake assay showed that the Suc uptake rate in roots of hps1 plants was about 4-fold higher than in wild-type plants (Fig. 6C). (For a detailed procedure of measuring Suc uptake rate, see "Materials and Methods.") Similarly, the accumulation of Suc in both shoots and roots of hps1 seedlings was much higher than that of wild-type plants, especially in root tissues under P+ conditions (Fig. 6D). The high content of Suc in shoots of the hps1 mutant was probably due to the transport of Suc from Figure 3. Analyses of anthocyanin, chlorophyll, and starch contents in wild-type and hps1 seedlings. A, Morphological appearance of 14-dold wild-type (WT) and hps1 seedlings grown on MS P+ and P2 media. B, Anthocyanins were extracted from the shoots of 14-d-old wild-type and hps1 seedlings grown on MS P+ and P2 media. The contents of anthocyanin were determined spectrophotometrically and expressed as A 530 mg 21 fresh weight (FW). C, Chlorophylls and carotenes were extracted from the shoots of 10-d-old wild-type and hps1 seedlings grown on MS P+ and P2 media. The contents of the pigments were determined spectrophotometrically. Asterisks indicate significant differences compared with wild-type plants (P , 0.05, t test). D, Starch accumulation in 11-d-old wild-type and hps1 seedlings grown on P+ and P2 media as revealed by iodine staining. root tissues with water through the xylem. We noticed that the Suc levels in both shoot and root tissues of wild-type plants were also elevated under Pi starvation (Fig. 6D). Since Suc is readily metabolized into Glc and Fru as it enters the metabolic pathway, we measured the contents of these two monosaccharides in hps1 and wild-type plants. The amount of Glc and Fru in hps1 plants was less than half of that in the wild type, except in Pi-starved roots, where the Glc level was about 25% lower (Supplemental Fig. S6). To further confirm that the hps1 mutant phenotype was caused by overexpression of the SUC2 gene, we introduced the SUC2 gene into the AtPT2-LUC line under the control of a cauliflower mosaic virus 35S promoter. Ninety-two independent transgenic lines were generated, and 90% of the lines showed the hps1 mutant phenotype (i.e. superinduction of the AtPT2-LUC gene, strong inhibition of primary root growth, formation of more lateral roots; Fig. 7A). In fact, most of these transgenic lines showed a stronger phenotype than hps1, probably due to the higher SUC2 expression driven by the strong 35S promoter. In contrast, when grown on Murashige and Skoog (MS) Suc-free medium in both P+ and P2 conditions for 14 d, there was no obvious phenotypic difference between the wild type and the hps1 mutant in terms of AtPT2-LUC expression and seedling morphologies (Supplemental Fig. S7). Taken together, these results demonstrated that the hps1 mutant phenotype was indeed caused by the overexpression of SUC2 that resulted in enhanced uptake of Suc from the culture medium.
In the Arabidopsis genome, there are nine annotated Suc transporter genes, which can be divided into three groups based on their sequence similarity (Sauer, 2007). SUC2 is a plasma membrane-associated highaffinity Suc/proton symporter that is specifically expressed in phloem companion cells and root vascular tissues (Stadler and Sauer, 1996). SUC1 and SUC5 are also plasma membrane-associated Suc transporters and share high sequence homologies with SUC2. However, SUC1 is mainly expressed in floral tissue and developing seeds, whereas SUC5 expression is embryo specific. In the same subgroup as SUC2, SUC6 and SUC7 are annotated as pseudogenes, and SUC8 and SUC9 are expressed in floral tissues (Sauer et al., 2004). SUC3 and SUC4 belong to two different subgroups. The SUC4 protein is localized in tonoplast, which is not involved in the phloem transport of Suc from source to sink tissues (Endler et al., 2006). SUC3 has very low affinity to Suc and is mainly expressed in sieve elements in sink tissues (Barker et al., 2000). To determine if the overexpression of SUC genes with different functions has similar effects as SUC2, we introduced SUC1, SUC5, SUC3, and SUC4 genes into AtPT2-LUC plants under the control of the 35S promoter. As shown in Figure 7B, only the overexpression of SUC1 and SUC5 had a similar effect as SUC2 on AtPT2-LUC gene expression and root development. This was probably due to the overexpression of plasma membrane-associated high-affinity Suc transporters (SUC1, SUC2, and SUC5) leading to increased uptake of Suc from culture medium, whereas SUC3 and SUC4 were not able to fulfill this role. Thus, we believe that it is the level of Suc, rather than the activity of a particular Suc transporter, that regulates the extent of plant responses to Pi starvation. In fact, the soilgrown hps1 and 35S-SUC1, 35S-SUC2, and 35S-SUC5 plants did not show obvious phenotypic differences Plants were grown on MS P+ and P2 media for 9 d. A, The cellular Pi was measured using the Ames (1966) method. B to D, The element contents of phosphorus (B), iron (C), and zinc (D) were determined by ICP-OES. Asterisks indicate significant differences compared with the wild type (P , 0.05, t test). White bars, Wild-type plants; black bars, hps1 plants. DW, Dry weight; FW, fresh weight. compared with the wild type. This is probably due to the absence of a high concentration of Suc in soils for these plants to take up. This phenomenon further supported the notion that high Suc is critical for plants to display hypersensitive responses to Pi starvation.

Reduction of Suc in Roots Severely Inhibits Plant Responses to Pi Starvation
To provide further genetic evidence that the accumulation of Suc is critical for plant responses to Pi starvation, we identified a Salk line (SALK_087046), designated as suc2-5, in which the T-DNA is inserted into the first exon of the SUC2 gene ( ). On sugar-free medium, suc2-5 seedlings have very short roots (Supplemental Fig. S8A). To examine the induction of APase activity and PSI gene expression, the seeds of wild-type and suc2-5 plants were sown on MS P+ medium with Suc and grown for 5 d and then transferred to MS P2 medium with no added Suc for another 5 d. As shown in Figure 8A, the suc2-5 mutant showed almost no blue staining on the root surface compared with wild-type plants, indicating a lack of APase activity. This result is consistent with that previously reported for the pho3 mutant, which contains a point mutation in the SUC2 gene (Zakhleniuk et al., 2001). Expression of four PSI genes (AtPT2, ACP5, RNS1, and AtIPS1) was also analyzed. Induction of these genes in roots of suc2-5 plants was significantly reduced, especially for AtIPS1 (Fig. 8C). However, in shoot tissues, the expression of these PSI genes, except RNS1, was higher than that of the wild type (Fig. 8D). Similar result was obtained when the suc2-5 mutation was introduced into the AtPT2-GUS marker line through a genetic cross (i.e. GUS expression under Pi starvation was enhanced in shoot tissue and blocked in root tissue; Fig. 8B). These contrasting expression patterns of PSI genes could be correlated with altered accumulation patterns of Suc in suc2-5 plants. In an early study, Gottwald et al. (2000) had shown that the transport of Suc from shoot to root through the phloem is greatly blocked in suc2 mutants; thus, more Suc is accumulated in shoots and less in roots. These results strongly supported the notion that the level of Suc inside plant cells is a major determinant of plant responses to Pi starvation.

Suc Plays a Global Regulatory Role in Mediating Plant Responses to Pi Starvation
To determine the effects of the hps1 mutation (or the high accumulation of Suc) at the genomic level, we compared the gene expression profiles between hps1 and wild-type plants by microarray analysis. The mRNAs used for analysis were isolated from 9-d-old seedlings (the same batch of mRNAs used for realtime PCR analysis as shown in Fig. 2). All the comparisons were made against the gene expression level of wild-type plants on P+ medium. The corrected P value (P , 0.001) was used to select the genes for comparison, and only those genes whose expression was 2-fold higher or lower compared with that of wild-type plants on P+ medium were used for analysis. The overall changes in gene expression patterns are shown in Supplemental Figure S9. This study is mainly focused on the analysis of genes whose expression is up-regulated.
We first compared the gene expression profiles of wild-type plants under P+ and P2 conditions. Under our experimental conditions, expression of 325 genes was induced, and 133 genes were repressed when wild-type plants were grown under Pi starvation. Among these 325 induced genes were those eight PSI genes used as markers in real-time PCR analysis (Fig.   Figure 5. Comparisons of root morphologies between wild-type and hps1 seedlings. Wild-type (WT) and hps1 plants were grown under P+ and P2 conditions for 9 d on vertically oriented petri plates. A, Photographs are of representative wild-type and hps1 whole seedlings grown on MS P+ and P2 media. B and C, Microscopic images of root middle parts (B) and root tips (C) with intact root hair from plants grown on MS P+ and P2 media. Table I; Supplemental Table S2). Pi starvation-induced genes included key Pi signaling components such as AtIPS1, At4, and a SPX domain-containing protein that play important roles in regulating Pi homeostasis (Burleigh and Harrison, 1999;Martín et al., 2000;Franco-Zorrilla et al., 2007). The 1,2-diacylglycerol 3-b-galactosyltransferase genes thought to be involved in the synthesis of galactolipid were also strongly induced. Galactolipids are shown to replace phospholipids in biomembranes under Pi starvation (Dö rmann and Benning, 2002). Another interesting observation is that 15 peroxidase genes are highly induced by Pi starvation (Supplemental Table S2). These may be involved in scavenging ROS accumulated during Pi deficiency.

2;
When comparing the gene expression profiles between wild-type and hps1 plants under normal growth conditions, we found that in hps1 mutants, there are 1,608 genes induced (Supplemental Table S1) and 1,071 genes repressed. Among the 1,608 induced genes in hps1 plants, 238 genes could also be induced by Pi starvation in wild-type plants (Supplemental Fig. S9; Supplemental Table S2). In other words, elevated Suc level alone was sufficient to induce 73% (238 out of 325) of PSI genes observed in wild-type plants. Table I lists some PSI genes induced in hps1 plants when grown in Pi-sufficient medium. AtIPS1, At4, and SPX3 are important regulators for Pi homeostasis, and they were among the highly induced genes in hps1 plants ( Table I), suggesting that these signaling components act downstream of Suc in regulating plant responses to Pi starvation. Besides those genes, others that are directly involved in Pi transport (AtPT1 and AtPT2), mobilization (ACP5 and RNS1), and synthesis of galactolipids (1,2-diacylglycerol 3-b-galactosyltransferase) were also induced in hps1 plants (Table I;  Supplemental Table S1). The induction of these PSI genes by elevated Suc level could be expected through a Suc-specific signaling pathway or by altered carbon metabolism. In either case, Suc seems to act at a very early step to trigger PSI gene expression.
We also compared the gene expression patterns of wild-type and hps1 plants on P2 medium. In contrast to wild-type plants on P2 medium, in which only 328 genes were induced, there were 3,678 genes induced in hps1 plants (Supplemental Fig. S9). Among these 3,678 genes, 305 genes overlapped with 328 genes induced in wild-type plants under Pi deficiency. Interestingly, 57% of the 328 PSI genes induced in wild-type plants were hyperinduced in hps1 plants (Table I; Supplemental Table S3). For example, on P2 medium, a hypothetical protein gene was induced 167-fold in the wild type but 2,706-fold in hps1; similarly, an APase gene (PAP25; At4g36350) was induced 101-fold in wild-type plants but 2,596-fold in hps1. AtIPS1, At4, and SPX domain-containing protein were also among these hyperinduced genes. In fact, AtIPS1 was the highest induced gene in hsp1 plants on P2 medium. The induction of the 1,2-diacylglycerol 3-b-galactosyltransferase gene had increased significantly in hps1 plants compared with the wild type (230.9-fold versus 20.5fold), suggesting that an active synthesis of galactolipids might take place in hps1 plants. These results extended our observations on the effects of hps1 mutation (or high Suc) on the expression of PSI genes (Fig. 2). Relative expression levels of SUC2 in 9-d-old wild-type (WT) and hps1 seedlings. C, Suc uptake rates of wild-type and hps1 plants. Wild-type and hps1 plants were grown on MS medium for 3 weeks, and the roots were immersed in the uptake solution containing [ 14 C]Suc and incubated for 2 h. The amount of [ 14 C]Suc uptake was determined by a scintillation counter. FW, Fresh weight. D, Soluble sugars were extracted from 9-d-old wild-type and hps1 seedlings grown on MS P+ and MS P2 media, and Suc contents were determined by gas chromatography-mass spectrometry. White bars, Wild-type plants; black bars, hps1 plants. Asterisks indicate significant differences compared with the wild type (P , 0.05, t test).
The strong enhancing effect of high Suc on PSI gene expression may come from the interaction between Suc and Pi. In fact, we found that there existed both synergistic and antagonistic interactions between Suc and Pi starvation signals. For example, the AtIPS1 gene was induced to 80-fold in hps1 plants on P+ medium and 167-fold in wild-type plants on P2 medium; however, in hps1 plants grown on P2 medium, this gene was induced by 2,706-fold (Table I). Similarly, a purple acid phosphatase (PAP25) gene was induced to 101-and 13-fold, respectively, by Pi starvation and hps1 mutation (or high Suc), but when hsp1 was Pi starved, the induction reached 2,596-fold (Table  I). In contrast, a putative peroxidase gene (At1g49570) was induced 19-and 46-fold by Suc and Pi starvation individually, but when both treatments were combined, the induction was only 14-fold, thus suggesting an antagonistic interaction. In summary, we have found 78 genes whose expression was synergistically induced by hps1 mutation and Pi starvation. And Suc and low Pi together had antagonistic effects on the expression of 60 genes (Supplemental Table S4). These results indicated that elevated levels of Suc had broader implications on Pi starvation-induced gene expression in plants.

DISCUSSION
When plants are subjected to Pi starvation, they display numerous adaptations to cope with the stress. There is growing evidence that Suc plays an important role in PSI responses (Zakhleniuk et al., 2001;Liu et al., 2005;Karthikeyan et al., 2007). Several reports, including this work, have shown that Pi starvation led to increased accumulation of Suc in plant leaves (Hammond and White, 2008). Increased Suc levels in leaves of Pi-deficient plants and activation of PSI genes have provided a tangible link between Suc and Pi (Liu et al., 2005;Karthikeyan et al., 2007). This work defines a broader role for Suc in many of the observed Pi starvation responses in Arabidopsis and tries to explore how Suc acts at the molecular level in regulating these responses through genetic and genomic analyses of a Suc overaccumulation mutant, hps1.
hps1 was identified as an Arabidopsis mutant that exhibited hyperinduction of PSI marker genes under Pi-deficient conditions. The products of these PSI marker genes are thought to be directly involved in Pi transport (AtPT1, AtPT2, Pht1;5), mobilization (ACP5 and RNS1), and control of Pi distribution between shoots and roots (At4, AtIPS1, and miRNA399), Figure 7. Comparison of AtPT2-LUC expression and plant morphology of wild-type, hps1, and transgenic lines overexpressing various SUC genes. A, Nine-day-old seedlings of wild-type (WT), hps1, and two transgenic lines expressing 35S-SUC2 grown on MS P+ (left panels) and P2 (right panels) media were sprayed with luciferins, and the images were taken with a CCD camera. B, Nine-day-old seedlings of wild-type, hps1, and transgenic lines overexpressing various SUC genes were sprayed with luciferins, and the images were taken with a CCD camera. Top panels show LUC activity, and bottom panels show plant morphology.
which are critical for plants to maintain Pi homeostasis under Pi deficiency conditions. Further comparison of expression profiles between wild-type and hps1 plants at the genomic scale revealed that, among 325 Pi starvation-induced genes found in the wild type, the induction of 190 genes (about 60%) was greatly enhanced in hps1 plants (Supplemental Table S3). These results indicated that mutation of the HPS1 gene has a global impact on PSI gene expression. Molecular and genetic analyses indicated that the mutant phenotype is caused by the overexpression of the SUC2 gene, which resulted in high accumulation of Suc in both shoot and root tissues by enhanced Suc uptake from the medium. This mutant offered genetic evidence that Suc is an important component for regulating PSI gene expression. Previous studies have shown that change in photosynthate level had a strong effect on PSI gene expression in white lupin and Arabidopsis (Liu et al., 2005(Liu et al., , 2010bKarthikeyan et al., 2007). This was demonstrated through stem girdling and photoperiod ma- Figure 8. Comparison of APase activity and PSI gene expression in wild-type and suc2-5 mutant plants. Wild-type and suc2-5 seedlings were grown on MS P+ medium with Suc for 5 d and then transferred to MS P2 medium with no added Suc for another 5 d. A, The APase activities on root surfaces of wild-type (WT) and suc2-5 seedlings grown on MS P2 medium as shown by BCIP staining. B, Histochemical analysis of AtPT2-GUS expression in wild-type and suc2-5 seedlings grown on MS P2 medium. C, Quantitative analysis of expression of four PSI genes in roots of wild-type and suc2-5 seedlings grown on MS P+ and MS P2 media. D, Quantitative analysis of expression of four PSI genes in shoots of wild-type and suc2-5 seedlings grown on MS P+ and MS P2 media. The PSI genes examined in C and D encode a high-affinity Pi transporter (AtPT2), an APase (ACP5), an RNase (RNS1), and a noncoding mRNA (AtIPS1). Asterisks indicate significant differences compared with the wild type (P , 0.05, t test). White bars, wild-type plants; black bars, suc2-5 plants.
nipulation. However, in these experiments, it was not possible to separate the role of Suc from other translocatable components acting as potential signaling molecules. In this work, the wild-type and hps1 plants were maintained at regular growth conditions (i.e. 16h-light/8-h-dark photoperiod with 100 mmol m 22 s 21 fluorescent light). Thus, this study provided more compelling evidence for the causal relationship between the elevated Suc level and enhanced PSI gene expression. Analyses of a SUC2 null allele, suc2-5, further support the notion that Suc is a major determinant of PSI gene expression. In the Arabidopsis genome, there are nine annotated Suc transporter genes (Sauer, 2007). However, SUC2 is the only Suc transporter that is specifically involved in phloem loading of Suc from leaf mesophyll cells (Sauer et al., 2004). It has been shown that the disruption of SUC2 function led to high accumulation of Suc in leaves and blocked Suc transport from leaves to root tissues (Gottwald et al., 2000). As a consequence, when suc2-5 is grown on Suc-free P2 medium, the expression of four PSI marker genes is enhanced in shoots. In contrast, the expression of these PSI genes and the induction of APase were largely attenuated in roots. Transfer of Suc from shoot to root seems to be a crucial part of signaling during Pi starvation, and Suc transporters have an important role in regulating sugar movement. Furthermore, the ectopic overexpression of SUC1 and SUC5, which have similar Suc transport properties as SUC2, can also recapitulate the hps1 mutant phenotype in wild-type plants ( Fig. 7; i.e. enhancing plant sensitivity to Pi starvation), demonstrating that it is the level of Suc, rather than a particular Suc transporter, that affects the magnitude of plant responses to Pi starvation.
In this work, we noticed that the expression levels of seven out of eight PSI genes examined by real-time PCR were largely enhanced in the hps1 mutant compared with the wild type even when they were grown under Pi-sufficient conditions (Fig. 2). The enhancing effect on the expression of some PSI genes (AtPT1, AtPT2, LePT1, LaSAP1, At4, and RNase2) by exogenously applied Suc was also reported by other groups in white lupin and Arabidopsis (Lejay et al., 2003;Karthikeyan et al., 2007). These results indicated that increased Suc level alone is sufficient to induce PSI gene expression independent of low-Pi stress. Furthermore, our genomic studies indicated that 73% (238 out of 325) of Pi starvation-induced genes can be directly induced by elevated levels of Suc in hps1 plants. These genes include some of the known regulatory components involved in Pi starvation responses and many downstream target genes that directly participate in Pi transport and mobilization. Although the elevated Suc level is sufficient to induce or enhance the expression of many PSI genes in Pi-sufficient conditions, the degree of its induction is much lower compared with that induced by Pi starvation. This observation implied that Suc is not just a simple link between low-Pi signal and PSI gene expression; it needs to interact with other factors under Pi starvation conditions to orchestrate the expression of a battery of PSI genes. This notion is further supported by the genomic data that 78 genes are synergistically induced by both Pi starvation and high levels of Suc, and the expression of 60 genes is antagonistically regulated (Table I; data not shown). Interestingly, the expression of AtIPS1, a key regulator of Pi homeostasis, is highest among the genes induced in Pi-starved hps1 plants. Expression of this gene is also synergistically induced by the combined effects of Suc and low-Pi signals (Table I).
These synergistically induced genes could be the converging points for Suc and Pi signals to regulate multiple plant responses to Pi starvation. The synergistic interaction of Suc and low-Pi signals in regulating gene expression at the genomic level has also been observed in a previous study using exogenously applied Suc to Pi-starved excised leaves (Mü ller et al., 2007). Thus, Suc appears to act as a major regulator of the expression of genes that are involved in plant responses to Pi starvation. Suc also plays an important role in Pi starvationinduced lateral root formation and root hair development . Increased production of lateral roots and root hairs and inhibition of primary growth are the major morphological changes associated with Pi starvation, which are thought to enhance Pi acquisition efficiency. When Arabidopsis seedlings were grown on Suc-free Pi-deficient medium, the development of lateral roots and root hairs was greatly suppressed (Nacry et al., 2005;Jain et al., 2007;Karthikeyan et al., 2007), suggesting that Suc is required for such Pi starvation-induced morphological changes. Further experimental evidence showed that the role of Suc might be to enhance plant sensitivity to auxin, an important mediator of lateral root development (Casimiro et al., 2001;Jain et al., 2007). Using the hps1 mutant, we were able to show that elevated Suc level alone is sufficient to inhibit primary root growth and enhance lateral root formation and root hair development, even when plants are grown under Pi sufficiency (Fig. 5), a phenotype that mimics a constitutive root Pi starvation response. Besides regulating PSI gene expression and root architecture changes, this study also shows that Suc level has a great effect on multiple Pi starvation responses, including the production of APase and the accumulation of starch and anthocyanin. In hps1 mutants that accumulate high levels of Suc, the production of APases, starch, and anthocyanin is greatly enhanced even in P+ medium, which mimics the expression of PSI genes and root development. In contrast, the APase activity on the root surface is largely reduced in the suc2-5 mutant. The high accumulation of anthocyanin and starch in shoots, and the reduced production of APase in roots, were also observed in pho3, another mutant allele of the SUC2 gene (Zakhleniuk et al., 2001;Lloyd and Zakhleniuk, 2004). Taken together, these results show that the hps1 mutant has enhanced sensitivity in multiple plant responses to Pi starvation, and this is primarily due to increased levels of Suc, which acts as a global regulator of plant responses to Pi starvation.
It is becoming evident that when plants are under Pi starvation, Suc level increases in their leaf tissues, probably because of the changes of gene expression or enzymatic activities that are associated with carbohy-drate metabolism. In Arabidopsis, Suc is transported from leaf to root due to the action of SUC2. The transported Suc then acts as a signal to trigger the root Pi starvation responses in addition to meeting the nutrient demands of an expanding root system. Our results also demonstrated that the plant Pi starvation response triggered by Suc is not due to the sequestration of cellular Pi by elevated sugar level, since the Pi concentration in root tissues of hps1 plants is actually higher than that of wild-type plants under both Pisufficient and Pi-deficient conditions (Fig. 4). Since Suc will be broken down to Glc and Fru when it enters the metabolic pathway, it could be argued that it may not be the real signaling molecule for eliciting the plant Pi starvation response. This study showed that the levels of Glc and Fru in hps1 plants are lower than in wildtype plants (Supplemental Fig. S6), excluding the possibility that enhanced plant Pi sensitivity in hps1 is caused by an increased level of Glc or Fru. These results support the hypothesis that a Suc-specific signaling pathway exists in triggering plant responses to Pi starvation.
In summary, our work demonstrates that Suc is a global regulator of plant responses to Pi starvation. Through genomic analyses, we showed that elevated levels of Suc can directly alter the expression of a large number of PSI genes that are involved in Pi signaling, transport, mobilization, and allocation between shoots and roots, which will ultimately help plants maintain Pi homeostasis and better adapt to this nutritional stress. A challenging task ahead is to identify signaling components that are the direct targets of Suc and understand how this carbohydrate signal is perceived and transmitted to elicit plant Pi responses at the molecular level. The data generated from our comparative genomic study will provide a valuable resource to search for these Suc-regulated signaling components. Alternatively, screening for genetic suppressors of hps1 and cloning the corresponding genes may also be an effective way to achieve this goal.

Plant Materials and Growth Medium
All Arabidopsis (Arabidopsis thaliana) plants used in this study, including mutants and transgenic plants, were in the Columbia background. Seeds were surface sterilized in 20% (v/v) household bleach solutions and washed with distilled water three times. Then they were planted on 8-cm-diameter sterile plates containing full-strength medium (MS P+) supplemented with 1% Suc and 1.2% agar or MS P2 medium (substituting 0.6 mM K 2 SO 4 for 1.2 mM KH 2 PO 4 ). After being cold stratified at 4°C for 2 d, these plates were placed vertically in the growth room. The growth room conditions were 24°C, 16-h-light/8-h-dark photoperiod, with 100 mmol m 22 s 21 fluorescent light.

Isolation of Mutants
T2 seeds were plated directly on MS P2 medium plates. After 8 d in the growth room, 100 mM luciferin in 0.1% Triton X-100 was uniformly sprayed on the seedling surface, and fluorescence images were captured by exposing the seedlings for 5 min on a highly sensitive iXon CCD camera (Andor Technol-ogy). Seedlings with altered LUC expression compared with wild-type plants were selected as candidates and transferred to soil. Seeds harvested from the putative mutants were retested for inheritance of the observed phenotype. The hps1 mutant presented in this work was backcrossed to wild-type plants before further characterization.

Analysis of APase Activity
In vivo APase assay was performed on 9-d-old seedlings vertically grown on P+ or P2 medium as described by Zakhleniuk et al. (2001). Agar (0.5%, w/v) containing 0.01% (w/v) BCIP was evenly overlaid on the root surface. After 8 h of color development, photographs were taken with a camera attached to a stereomicroscope (Olympus SZ61).
For in-gel APase assay, about 50 mg of seedlings was ground in liquid N 2 and total protein was extracted using extraction buffer (0.1 M KAc, 20 mM CaCl 2 , 2 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride). The same amount of protein was separated on a 10% SDS-PAGE gel. After electrophoresis, the gel was washed in distilled, deionized water at 4°C six times (10 min each) to renature the proteins and equilibrated twice with sodium acetate buffer (50 mM sodium acetate and 10 mM MgCl 2 , pH 4.9). Then, the gel was stained with 0.5 mg mL 21 Fast Black K coupled with 0.3 mg mL 21 b-naphthyl phosphate dissolved in sodium acetate buffer at 37°C for 4 h (Trull and Deikman, 1998;Zakhleniuk et al., 2001).
Quantitative analysis of APase activity was done by mixing 10 mL of protein extract with 470 mL of MES buffer (15 mM MES, pH 5.5, and 0.5 mM CaCl 2 ) and 20 mL of p-nitrophenyl phosphate, disodium salt. After incubation at 27°C for 30 min, the reaction was terminated with 500 mL of 0.25 M NaOH. The developed color was determined spectrophotometrically at 412 nm, and APase activity was expressed as A 412 mg 21 protein (Richardson et al., 2001).

Quantitative Measurements of Anthocyanin, Chlorophyll, and Inorganic Pi
Anthocyanin was extracted with 1% HCl in methanol from shoots of the seedlings grown on P+ or P2 medium for 9 d. Anthocyanin content was measured as described previously (Mita et al., 1997). Chlorophyll was extracted with 80% acetone from leaves of the seedlings grown on P+ or P2 medium for 9 d. The chlorophyll a, chlorophyll b, and carotene contents were measured as described previously (Arnon, 1949). Inorganic Pi was determined using the method described by Ames (1966).

Starch Analysis
For starch analysis, 11-d-old seedlings were immersed in 96% (v/v) ethanol overnight to remove pigments and stained with 1% iodine for 30 min. After a brief wash in distilled, deionized water, the samples were kept in 96% ethanol, and starch accumulation was observed with a stereomicroscope.

Histochemical Analysis of GUS Activity
For histochemical analysis of GUS activity, whole seedlings were incubated for 6 h at 37°C in GUS staining solution (2 mM 5-bromo-4-chloro-3-indolyl-b-glucuronic acid in 50 mM sodium Pi buffer, pH 7.2) containing 0.1% Triton X-100, 2 mM K 4 Fe(CN) 6 , 2 mM K 3 Fe(CN) 6 , and 10 mM EDTA. The stained seedlings were then transferred to 50%, 70%, and 100% (v/v) ethanol to remove the chlorophyll. The whole seedlings were photographed with a stereomicroscope (Olympus). The images of the shoot-root junctions and root tips were taken with a differential interference contrast microscope (Olympus).

Inductively Coupled Plasma-Optical Emission Spectrometry Elemental Profiling
Nine-day-old seedlings were cut into shoot and root parts and dried at 80°C for 2 d. The tissue was digested for 8 min at 120°C and for 15 min at 160°C with concentrated HNO 3 and diluted with distilled, deionized water. The contents of phosphorus, sodium, magnesium, potassium, calcium, zinc, iron, manganese, and sulfur were determined by inductively coupled plasmaoptical emission spectrometry (ICP-OES; IRIS Intrepid II XSP; Thermo Electron).

Generation of Transgenic Plants
The genomic sequences of SUC1, SUC2, SUC4, and SUC5 were amplified from wild-type Arabidopsis plants. The sequence of SUC3 was amplified by RT-PCR from a wild-type plant. The PCR products were cloned into the vector pEASY-Blunt (Transgen) and sequenced. Then, they were subcloned into the plant T-DNA vector pCAMBIA1301 under the control of the cauliflower mosaic virus 35S promoter. The vector has a hygromycin-resistant gene as a selectable marker for plant transformation. These constructs were then transformed into wild-type plants by the floral dip method (Clough and Bent, 1998).

Quantitative Real-Time PCR Analysis
Total RNA was extracted with the Qiagen RNAeasy kit. One microgram of DNase-treated RNA was reverse transcribed in a 20-mL reaction using reverse transcriptase (Toyobo) according to the manufacturer's manual. cDNA was amplified using SYBR Premix Ex Taq (TaKaRa) on the Bio-Rad CFX96 realtime PCR detection system. Ubiquitin-conjugating enzyme 21 mRNA was used as an internal control. The genes and their primers are listed in Supplemental Table S5.

Quantitative Analysis of Sugar Contents
Nine-day-old seedlings were cut into shoot and root parts and ground in liquid nitrogen thoroughly. Soluble sugar was extracted in 80% (v/v) ethanol twice at 80°C for 15 min. Samples were centrifuged at 4,000 rpm for 10 min and then dried at 85°C. For methoximation, 80 ml of methoxyamine hydrochloride in pyridine (20 mg mL 21 ) was added at 30°C for 90 min. After, 80 ml of N-methyl-N-trimethylsily-trifluoroacetamide was added, and the mixture was incubated at 37°C for 30 min. The sugar derivatives were analyzed by gas chromatography-mass spectrometry on an Agilent 7890A GC/5975 mass spectrometer using a DB-5ms column (Agilent Technologies). A temperature program was implemented as follows: initially at 180°C, followed by heating to 300°C at 10°C min 21 , and then held at 300°C for a further 10 min. Myoinositol was used as an internal standard.

Suc Uptake
Three-week-old seedlings grown on MS medium without supplemented Suc were used for Suc uptake experiments. After incubation in pretreatment solution (MS liquid medium, pH 5.7) for 30 min, the roots were incubated in the uptake solution (MS medium with 0.1% Suc, pH 5.7) containing [ 14 C]Suc (0.5 mCi mL 21 ) and incubated for 2 h. After two washes in desorption solution (MS liquid medium and 1% Suc, pH 5.7), the seedlings were placed in scintillation vials and 3 mL of scintillation cocktail was added. The amount of [ 14 C]Suc uptake was assessed on a scintillation counter and expressed as cpm mg 21 fresh weight.

Microarray Analysis
The microarray experiments were performed at ShanghaiBio Corporation using the Agilent Arabidopsis microarray analysis platform. Total RNA was isolated from three replicates of wild-type and hps1 seedlings grown on P+ and P2 media for 9 d using the RNase kit (Qiagen). First-and second-strand cDNA were generated using Moloney murine leukemia virus reverse transcriptase with T7 oligo(dT) 24 primer. Copy RNA was in vitro synthesized by T7 RNA polymerase and labeled with Cyanine 3 (Agilent). After purification and fragmentation, copy RNA was hybridized to the slides for 17 h at 65°C using the Agilent hybridization kit according to the manufacturer's instructions. The gene probes on the chips are synthesized according to the cDNA sequences released in The Arabidopsis Information Resource 8 (2008) The Arabidopsis Genome Initiative locus numbers for the major genes discussed in this article are as follows: SUC1 (AT1G71880), SUC2 (AT1G22710), SUC3 (AT2G02860), SUC4 (AT1G09960), SUC5 (AT1G71890), At4 (At5G03545), AtIPS1 (At3G09922), AtPT1 (At5G43350), AtPT2 (At2G38940), Pht1;5 (At2G32830), RNS1 (At2G02990), miR399D (AT2G34202), and ACP5 (AT3G17790).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Map of the pSuperTag2 plasmid.
Supplemental Figure S2. Comparison of AtPT1-GUS expression between wild-type and hps1 plants.
Supplemental Figure S3. APase activities in wild-type and hps1 plants.
Supplemental Figure S4. Comparison of ion contents between wild-type and hps1 plants.
Supplemental Figure S5. Comparison of the number of lateral roots produced in wild-type and hps1 mutant plants.
Supplemental Figure S6. Glc and Fru contents in wild-type and hps1 seedlings.
Supplemental Figure S7. AtPT2-LUC expression and morphologies of wild-type and hps1 seedlings grown on MS Suc-free medium.
Supplemental Figure S8. Phenotypic and molecular analyses of the suc2-5 mutant.
Supplemental Figure S9. Diagrams showing the number of genes up-and down-regulated by Pi starvation and hps1 mutation from microarray analysis.
Supplemental Table S1. Genes whose expression is up-and downregulated in hps1 plants grown on P+ medium.
Supplemental Table S2. Genes whose expression is up-and downregulated in wild-type plants grown on P2 medium.
Supplemental Table S3. Genes whose expression is up-and downregulated in hps1 plants grown on P2 medium.
Supplemental Table S4. Genes whose expression is up-and downregulated by both Pi starvation and hps1 mutation.
Supplemental Table S5. Primers used for real-time PCR analysis.