Characterization of a sub-family of Arabidopsis genes with the SPX domain reveals their diverse functions in plant tolerance to phosphorus starvation


(fax +86 571 88206617; e-mail


Four genes of Arabidopsis (At5g20150, At2g26660, At2g45130 and At5g15330) encoding no conservative region other than an SPX domain (SYG1, Pho81 and XPR1) were named AtSPX1–AtSPX4. The various subcellular localizations of their GFP fusion proteins implied function variations for the four genes. Phosphate starvation strongly induced expression of AtSPX1 and AtSPX3 with distinct dynamic patterns, while AtSPX2 was weakly induced and AtSPX4 was suppressed. Expression of the four AtSPX genes was reduced to different extents in the Arabidopsis phr1 and siz1 mutants under phosphate starvation, indicating that they are part of the phosphate-signaling network that involves SIZ1/PHR1. Over-expression of AtSPX1 increased the transcript levels of ACP5, RNS1 and PAP2 under both phosphate-sufficient and phosphate-deficient conditions, suggesting a potential transcriptional regulation role of AtSPX1 in response to phosphate starvation. Partial repression of AtSPX3 by RNA interference led to aggravated phosphate-deficiency symptoms, altered P allocation and enhanced expression of a subset od phosphate-responsive genes including AtSPX1. Our results indicate that both AtSPX1 and AtSPX3 play positive roles in plant adaptation to phosphate starvation, and AtSPX3 may have a negative feedback regulatory role in AtSPX1 response to phosphate starvation.


The essential macronutrient phosphorus (P) serves as a structural element for organic compounds, an intermediate for bio-energetic processes, and a component in signaling cascades (Poirier and Bucher, 2002). The chemical properties of P determine its low bioavailability in soils. To cope with low nutritional phosphate (Pi) availability, plants have evolved sophisticated metabolic and developmental strategies to facilitate Pi acquisition and remobilization. Plants may increase Pi uptake by expanding their root system to achieve a higher absorption surface, and enhance Pi uptake efficiency by organic anion excretion and elevated expression of transporters and acid phosphatase (Abel et al., 2002). Although knowledge of Pi uptake and the Pi-signaling system has increased recently (Bari et al., 2006; Chen et al., 2007; Chiou et al., 2006; Devaiah et al., 2007a,b; Fujii et al., 2005; Shin et al., 2006), the molecular network underlying plant adaptation to Pi starvation is still to be elucidated.

The hydrophilic and poorly conserved SPX domain (SYG1/Pho81/XPR1) is found at the N-termini of various proteins, particularly signal transduction proteins (Barabote et al., 2006). There are genes harboring the SPX domain in all major eukaryotic kingdoms. SYG1 inhibits transduction of the mating pheromone signal in yeast via direct binding to the G-protein β-subunit (Spain et al., 1995). PHO81, the putative sensor of Pi levels in yeast, is a cyclin-dependent kinase (CDK) inhibitor that is induced by Pi starvation, interacting with cyclin PHO80 to repress the activity of CDK PHO85, thus promoting the expression of PHO2/PHO4 and enhancing yeast tolerance to Pi starvation (Lenburg and O’Shea, 1996). The human protein XPR1 functions as a cell-surface receptor for xenotropic and polytropic retroviruses, and is probably also involved in G-protein-mediated signaling (Battini et al., 1999).

Most identified plant SPX gene products are involved in responses to environmental cues or internal regulation of nutrition homeostasis. Barley IDS4 (iron-deficiency specific clone 4) contains part of the SPX domain and is preferentially expressed in Fe-deficient roots (Nakanishi et al., 1993). Arabidopsis PHO1, harboring both SPX and EXS domains, plays a role in loading root Pi into the xylem vessels, and loss of PHO1 function in pho1 mutants results in Pi deficiency in above-ground tissues (Hamburger et al., 2002; Poirier et al., 1991). A PHO1 homolog in Arabidopsis may have a similar role in Pi loading and signaling (Wang et al., 2004). The product of the tomato IDS4-like gene interacts with the leucine zipper domain of a hypoxia-induced transcription factor involved in the low-oxygen response (Sell and Hehl, 2005). Homologous to yeast SYG1, the Arabidopsis SHORT HYPOCOTYL UNDER BLUE 1 (SHB1) protein acts in cryptochrome signaling and possible phytochrome-mediated light responses (Kang and Ni, 2006).

The Arabidopsis genome encodes 20 genes with the SPX domain, grouped into four sub-families. Three sub-families, with a total of 16 members, encode proteins with the SPX domain and an extra conservative domain, including AtPHO1, which is involved in regulation of Pi homeostasis in Arabidopsis (Wang et al., 2004). The other four members (At5g20150, At2g26660, At2g45130 and At5g15330) form a unique sub-family encoding proteins of about 300 amino acids in size, with no conservative region of the Pfam-A type other than the SPX domain. Here these four genes are named sequentially as AtSPX1AtSXP4. To probe their potential role in plant responses to Pi signaling, we characterized their expression profiles and localization of their GFP fusions. Intriguing differences were revealed in their dynamic expression patterns under Pi deficiency and in the localization of the GFP fusions. Relevant over-expression materials and T-DNA insertion mutants or RNAi materials were subsequently generated, and phenotypes were investigated under Pi deficiency or Pi sufficiency. Gene expression patterns and plant developmental phenotypes in the corresponding genetic materials highlighted the evolutionary diversification in their biological roles. Our results suggest involvement of AtSPX1 in transcriptional regulation of Pi-responsive genes, and of AtSPX3 in potential negative feedback regulation of the Pi-signaling network.


Structural and homologous characterization of four SPX domain genes

The four AtSPX genes encoding no conserved region except for a SPX domain are referred to as AtSPX1 (At5g20150), AtSPX2 (At2g26660), AtSPX3 (At2g45130) and AtSPX4 (At5g15330). The AtSPX genes all have three exons and two introns (Figure 1a). The position and size of the three exons are fairly well conserved, with the SPX coding region mainly spanning the first two exons. Pairwise comparison reveals that AtSPX1 and AtSPX2 share more structural features, while AtSPX3 is similar to AtSPX4. The SPX domain in AtSPX1 and AtSPX2 spans the first two exons to the 5′ end of the third exon. Thus, the greatest divergence in the structure of the four genes is the relative lengths of the intron and exon ranges spanned by the SPX domain.

Figure 1.

 Structure and homology of AtSPX1–4.
(a) Intron (white box) and exon (black box) organization of AtSPX1–4. The numbers indicate the length in base pairs (bp) of the corresponding exon.
(b) Modular structure of Arabidopsis proteins encompassing a SPX domain. The numbers indicate the amino acid ranges of the SPX domain (Pfam-A) spanned. Pfam-B, less conservative; unknown, non-conservative.
(c) Phylogenetic tree of AtSPX1–4 and homologous proteins from other eukaryotes. In addition to AtSPX1–4, other Arabidopsis SPX proteins shown are: AtPHO1 (At3g23430) with an extra EXS domain, At1g63010 with an extra transporter-like domain (MFS_1), and At1g02860 with a zinc finger. Other proteins include tomato (Le) IDS4-like, barley HvIds4, cotton (Ag) ADL143Wp and yeast (Sc) PHO81, as well as six rice SPX proteins AK072067, AK107245, AK108539, XM-470931, XM-479156 and AK066364. The tree was created using the ClustalX program ( (Jeanmougin et al., 1998).

Their deduced proteins are similar lengths (245–318 amino acids) with 41–72% similarity (Figure 1b). Analysis of their tripartite SPX domain revealed 50–77% similarity of amino acids in the entire SPX domain, but >90% in each sub-domain, with the third sub-domain being the most conserved (Figure S1). Representative genes that are homologous to the AtSPX genes are shown in Figure 1(c), and their amino acid sequence alignment analysis is shown in Figure S2. AtSPX1–4 are more homologous to SPX genes from other plant species that to other Arabidopsis SPX genes.

Subcellular localization of the AtSPX proteins

No signal peptide has been predicted for the deduced AtSPX proteins, except for a bipartite nuclear-targeting sequence KRLKLIGSKTADRPVKR from amino acids 30–46 in AtSPX1 ( C-terminal GFP fusions driven by CaMV 35S were used for visualization of AtSPX localization. Expression clones were transformed into onion epidermal cells for transient assay (Figure 2a,c,e,g). AtSPX1–GFP and AtSPX2–GFP were localized exclusively to the nucleus, while fluorescence of AtSPX4–GFP was detected in regions that included the cytoplasmic membrane. AtSPX3–GFP was present in onion cells as many fluorescent speckles in the cytoplasm. The peroxisome-localized RFP plasmid pCAMB-RFPPTS1 was co-transformed into onion cells with the AtSPX3–GFP plasmid, and partial overlapping of green and red fluorescence was observed (Figure S2).

Figure 2.

 Expression of CaMV 35S::AtSPX–GFP fusion genes in onion and Arabidopsis.
Expression of AtSPX1–GFP in onion cells (a) and Arabidopsis mesophyll protoplasts (b), of AtSPX2–GFP in onion cells (c) and Arabidopsis root (d), of AtSPX3–GFP in onion cells (e) and Arabidopsis mesophyll protoplasts (f), and of AtSPX4–GFP in onion cells (g) and Arabidopsis root (h). The red in (b) and (f) is auto-fluorescence of chlorophyll. The lower part of each picture is the merged image or bright-field image. Bar = 50 μm.

Transgenic Arabidopsis materials were subsequently generated for the AtSPX–GFP fusion proteins. Stable expression of AtSPX2–GFP in Arabidopsis roots was localized to the same cellular location as in the transient assay (Figure 2d). Fluorescence of AtSPX4–GFP was found in the cytoplasmic membrane, and weak signals were observed in some intercellular membranous systems (Figure 2h). There was no obvious green fluorescence in transgenic seedlings harboring AtSPX1–GFP or AtSPX3–GFP constructs. Therefore, a transient expression assay was performed in Arabidopsis mesophyll protoplasts. AtSPX1–GFP was localized to the nuclei in Arabidopsis protoplasts as for onion cells (Figure 2b), consistent with the prediction by PROSITE. Fluorescence of AtSPX3–GFP partially overlapped with the auto-fluorescence of chlorophyll in the mesophyll protoplasts (Figure 2f). AtSPX3 may be localized in intercellular membranous particles, including microbodies and chloroplasts.

Histochemical analysis for the AtSPX genes

To study the expression profiles of the AtSPX genes, the 1.5–2.0 kb promoter/5′ UTR region for each gene was amplified (designated AtSPX1p, AtSPX2p, AtSPX3p and AtSPX4p) and fused to β-glucuronidase (GUS) in pCambia1391Z and then transformed into Arabidopsis (Col).

Continuous GUS staining in young seedlings from germination revealed varied patterns (Figure 3a,b). AtSPX1p and AtSPX4p showed strong staining in cotyledons and hypocotyls, while AtSPX2p showed moderate staining in radicles. AtSPX3p showed no activity during early germination, and the first staining observed was in hypocotyls 5 days after germination (DAG). The full opening of cotyledons was followed by decreased promoter activity in cotyledons and radicles, but increased in hypocotyls for AtSPX1p, AtSPX12p and AtSPX14p. Later, there was obvious staining for AtSPX1p in leaf veins, preferentially in the abaxial region of leaf blades, and also in the branching regions of primary and lateral roots (Figure 3b). AtSPX2p was active in cotyledons and roots on appearance of the first leaf. AtSPX3p–GUS staining was only observed in leaf veins and shoot apical meristems but not in root systems under optimal nutrition. AtSPX4p activity was high in shoot apical meristems and weak in lateral root initiation regions (vascular tissue).

Figure 3.

 GUS staining of transgenic Arabidopsis harboring AtSPX promoter–GUS fusions.
GUS fusions of AtSPX1–4 promoters are marked ‘1p’–‘4p’, respectively. (a) From germination to 4 DAG. Bar = 0.5 mm for seeds, 1 mm for young seedlings. (b) From 7–14 DAG: distinct activities are seen in root, leaf veins and the shoot apical meristem. Bar = 1 mm. (c) Mature plants: various staining patterns in leaf and stem. Bar = 1 mm. (d) GUS staining under high Pi (HP, 1 mm NH4H2PO4) or low Pi (LP, 15 μm NH4H2PO4). AtSPX1p was induced by Pi deficiency at 3 and 7 DAG; AtSPX2p was moderately induced at 7 DAG; AtSPX3p was intensively induced at 7 DAG. Bar = 3 mm.

In mature plants, AtSPX1p–GUS staining was observed in rosette leaves (veins and trichomes) and cauline leaves (Figure 3c). AtSPX2p–GUS activity was not found in leaves, but was seen in stems at branching of axillary shoots. Staining for AtSPX3p–GUS was only observed in the abaxial region of the leaf blade. AtSPX4p activity was evident in the main vein of rosette leaves. All promoters uniformly displayed activity in mature pollen grains, and the staining intensity was consistent with pollen activity following petal unfolding (Figure S3). Control plants transformed by pCambia1391Z without insertions showed weak GUS staining in the base of pistils, young siliques and flower stalks (data not shown).

Expression responses of the AtSPX genes to Pi starvation

To study the response of the AtSPX genes to Pi starvation, transgenic seeds harboring promoter–GUS fusions were germinated on media with various Pi concentrations. Pi starvation enhanced the expression of all AtSPXp–GUS fusions except AtSPX4p–GUS (Figure 3d). AtSPX1p activity was rapidly induced by Pi starvation, with detectable promotion at 3 DAG and to an even higher degree at 7 DAG. There were detectable increases in AtSPX2p and AtSPX3p activities at 5 DAG (data not shown), and AtSPX3p was most highly induced at 7 DAG.

Quantitative RT-PCR was subsequently performed (Figure 4a,b). Pi starvation for 1 day markedly increased AtSPX1 transcript abundance, especially in roots. Pi starvation for 5 days led to notable expression enhancement of AtSPX1–3, with higher levels in roots than shoots. Re-supply of Pi for 1 day quickly decreased their abundance, especially of AtSPX3 in roots. In contrast, Pi starvation for 1 day reduced the level of AtSPX4 transcript to half that before treatment. Prolonged Pi deficiency did not affect AtSPX4 transcription further, but re-supply of Pi for 1 day resulted in complete recovery. The results of semi-quantitative RT-PCR distinguished AtSPX1 from AtSPX3 (Figure 4c). AtSPX1 was induced by one-week starvation on 200 μm Pi, and AtSPX3 was induced by one week on 70 μm Pi. AtSPX1 also responded more rapidly, with distinctly increased transcript levels after 12 h Pi starvation (2 μm Pi), while there was an obvious increase in AtSPX3 transcripts only after 4 days of Pi starvation.

Figure 4.

 Responses of AtSPX genes to Pi starvation in Col.
(a) Quantitative RT-PCR analysis of AtSPX1–3. Expression was relative to the transcript levels in shoots or roots under Pi-sufficient conditions (1 mm Pi). CK, Pi-sufficient; 1, 5 and 10 days, duration of Pi starvation (15 μm) (days); R1d, 10 days of Pi starvation followed by 1 day on Pi-sufficient substrate.
(b) Quantitative RT-PCR analysis of AtSPX4. Other conditions are same as in (a).
(c) Semi-quantitative RT-PCR analysis of AtSPX1 and AtSPX3. The dosage effects (1 week on media with varied Pi concentration, μm) and the dynamic influences of Pi starvation (2 μm Pi) on gene expression are shown in the upper and lower panels, respectively.

Transgenic plant analysis

For functional characterization, generation of over-expression materials for the AtSPX genes was paralleled by isolation of mutants or RNAi materials. Expression of the Pi starvation-inducible genes ACP5, PAP2 and RNS1 was enhanced in AtSPX1-over-expressing plants relative to the wild-type (WT). This expression enhancement was more significant under Pi sufficiency (Figure 5a,b) than under Pi deficiency. The 3–5-fold increased AtSPX1 expression was paralleled by similar elevations of ACP5 and PAP2 expression in shoots, and a greater increase of ACP5 expression in roots under Pi sufficiency. Under Pi starvation, AtSPX1 expression was not so obviously changed, but the other three genes were clearly upregulated in shoots, and the expression of RNS1 was clearly increased in roots. Together, these results indicate that AtSPX1 may be actively involved in Pi-signaling pathways by regulating the expression of the Pi-responsive genes ACP5, PAP2 and RNS1.

Figure 5.

 Overexpression of AtSPX1 resulted in altered expression of Pi-responsive genes.
(a) Quantitative RT-PCR analysis of Pi starvation-responsive genes in Col, AtSPX1-over-expressing lines (SPX1-O1 and SPX1-O2) and an AtSPX1 mutant (SALK_039445), under Pi-sufficiency conditions (1 mm) at 10 DAG.
(b) Expression profiles of the same genes under Pi-starvation conditions (15 μm) at 10 DAG.

The T-DNA insertion mutants SALK_039445, SALK_080503 and SALK_019826, corresponding to interruption of AtSPX1, AtSPX2 and AtSPX4, respectively, were obtained from the Arabidopsis Biological Resource Center and propagated and validated for exact genome insertion and complete expression deletion. There were no obvious phenotypes for these mutants under either Pi sufficiency or deficiency. Because a T-DNA insertion mutant of SPX3 was not available, an RNAi strategy was adopted and two AtSPX3-repressed lines were developed (Figure 6a). The AtSPX3 RNAi plants presented no detectable phenotype under Pi sufficiency.AtSPX1, AtSPX2 and AtSPX4 were not significantly affected in AtSPX3-repressed lines under Pi sufficiency, suggesting specific interference on AtSPX3 (Figure S4).

Figure 6.

 Repression of AtSPX3 via RNAi interfering led to hypersensitive response to Pi-starvation.
(a) Semi-quantitative RT-PCR analysis of AtSPX3 in Col and transgenic plants. Plants grown under Pi deficiency (15 μm Pi) to 11 DAG were used for RNA analysis. Upper panel, 32 cycles for AtSPX3 amplification; lower panel, 28 cycles for Actin amplification.
(b) Growth status of Col and AtSPX3-repressed lines (Ri1 and Ri2) under Pi deficiency (15 μm Pi) at 11 DAG. Bar = 1 cm.
(c) The primary root length (PRL) of WT and AtSPX3-repressed lines under Pi sufficiency (1 mm Pi) and Pi deficiency (15 μm Pi).
(d) Quantitative RT-PCR analysis for Pi-starvation-induced genes in AtSPX3-repressed lines. The genes shown are AtSPX1, PAP2 (At1g66390), RNS1 (At2g02990), IPS1 (At3g09922), At4 (At5g03545), AtPht1;4 (At2g38940) and AtPht1;5 (At2g32830).

AtSPX3-repressed lines showed lowered plant tolerance to Pi starvation, including reduced root systems and impaired aerial parts (Figure 6b). The primary root elongation in AtSPX3-repressed lines was much reduced by Pi starvation compared to WT (Figure 6c). The concentration of P in the WT and AtSPX3-repressed lines were determined (Table 1). Total P (mg P per g dried plant) was significantly higher in shoots and lower in roots of the two lines Ri1 and Ri2 compared to that in WT. Pi concentration (mg Pi per g fresh plant) increased by >60% in the shoots of AtSPX3-repressed lines compared to WT. The Pi level in the roots was unchanged in Ri1, but increased in Ri2.

Table 1.   Total phosphorus (P) (mg g−1 DW) and inorganic phosphate (Pi) concentrations (mg g−1 FW) in AtSPX3-repressed lines (Ri1 and Ri2) and the wild-type (WT) grown on Pi-deficient medium (15 μm NH4H2PO4)
  1. Shoots and roots were sampled from 11-day-old seedlings. The values represent the means ± SE of two independent experiments (n = 8).

  2. aSignificant difference (< 0.01) in means between WT and transgenic plants.

PShoot17.65 ± 1.7219.72 ± 1.32a20.02 ± 0.75a
Root36.04 ± 3.0532.72 ± 2.47a28.02 ± 2.25a
PiShoot0.103 ± 0.020.166 ± 0.02a0.174 ± 0.03a
Root0.379 ± 0.060.404 ± 0.030.522 ± 0.04a

Relevant gene expression was analyzed to further understand the hypersensitive phenotype of the AtSPX3-repressed lines. The Pi-responsive genes examined comprise the TPS1/Mt4 family genes IPS1 and At4, the Pht1 transporter family (nine members), the secreted phosphatase ACP5 gene, ribonuclease genes RNS1 and RNS2, the anthocyanin synthesis regulator PAP2 gene (production of anthocyanin pigment 2), and the other AtSPX genes. There were considerable alterations in expression of the genes (Figure 6d). Of the transporter family tested, only expression of Pht1;4 (AtPT2, Pht4) and Pht1;5 (Pht5) clearly increased in shoots. Similarly, PAP2 and RNS1 were markedly elevated in shoots of transgenic plants. AtSPX1 showed prominent expression enhancement in roots but a weak increase in shoots. IPS1 and At4 were upregulated in whole plants, more markedly in roots than shoots.

The AtSPX family is involved in Pi-signaling pathways controlled by PHR1 and SIZ1

To determine whether the SPXs are involved in Pi-signaling pathways, we further investigated the expression of AtSPX genes in mutants including pho1 (Wang et al., 2004), pho2 (Bari et al., 2006), phr1 (Rubio et al., 2001) and siz1 (Miura et al., 2005). Expression of all AtSPX genes under Pi starvation was not obviously changed in pho1 and pho2 (data omitted), but was clearly impaired in siz1 and phr1 compared to WT (Figure 7a). AtSPX1 was markedly repressed in siz1 and phr1. AtSPX2 expression was downregulated more in siz1 than in phr1. AtSPX4 expression was reduced to about half in siz1, with a mild decrease in phr1. AtSPX3 expression was not significantly reduced in siz1, but was markedly downregulated to <10% in phr1. Consistent results were obtained for AtSPX3p–GUS staining in WT and phr1 (Figure 7b), strongly suggesting that expression of AtSPX3 requires PHR1. These results suggest that AtSPX3 acts downstream of PHR1, while AtSPX1 and AtSPX2 have both PHR1 and SIZ1 as upstream regulators in Pi signaling.

Figure 7.

 The AtSPX family is involved in Pi-signaling pathways.
(a) Quantitative RT-PCR analysis of AtSPX genes in Col and mutants under Pi starvation. Whole seedlings from Pi-sufficient medium at 7 DAG were transferred to Pi-deficient medium (15 μm NH4H2PO4) for another one week, and then were harvested for RNA extraction.
(b) AtSPX3p–GUS staining in Col (left) and phr1 (right) under Pi starvation at 10 DAG. Bar = 5 mm.


The AtSPX genes are highly homologous to each other in structure and sequence, but a subcellular localization assay discriminated between them and hinted at functional diversification. Various expression patterns for these genes were observed in response to Pi starvation. This indicates that plant genes with the SPX domain have diverse functions.

The expression of AtSPX1 was most rapidly induced by Pi starvation. The nuclear localization of AtSPX1–GFP in Arabidopsis mesophyll protoplasts suggests its involvement in the transcriptional control of gene expression in response to Pi starvation. AtSPX1 over-expression increased the expression of the Pi-responsive genes ACP5, RNS1 and PAP2. The S-like ribonuclease RNS1 and the acid phosphatase ACP5 are involved in Pi recycling (Bariola et al., 1994; Olczak et al., 2003; del Pozo et al., 1999). The anthocyanin accumulation-enhancing gene PAP2 is involved in scavenging active oxygen species (Borevitz et al., 2000; Nagata et al., 2003). Therefore, AtSPX1 may be involved in transcriptional activation of genes related to P mobilization and scavenging of active oxygen species in response to Pi starvation. The expression levels of these Pi-starvation-inducible genes were not altered between the AtSPX1 knockout mutant and WT, indicating that induction of these genes by Pi starvation was not only controlled by AtSPX1, although data showed that AtSPX1 was under the control of AtPHR1. PHR1 is expected to have multiple functions in regulation of the Pi-signaling pathway.

RNAi of AtSPX3 results in a hypersensitive response to Pi starvation, followed by a significant decrease in total P allocation to roots and an increase of both total P and Pi in shoots. The AtSPX3-repressed plants were quite different from the reported Pi-starvation hypersensitive mutant PDR2, which shows systemic impaired growth (Ticconi et al., 2004). The Pi-signaling-related genes IPS1 (Martin et al., 2000) and At4 (Shin et al., 2006), together with AtSPX1, showed more enhanced expression in roots than in shoots. The Pi-recycling-related genes RNS1, Pht1;4 and Pht1;5 (Mudge et al., 2002) and the anthocyanin synthesis-related gene PAP2 were more increased in shoots than roots. These changes in gene expression were consistent with more severe P deficiency in roots and improved P remobilization in transgenic interfering plants compared to WT. This result implies that AtSPX3 has a positive role in plant adaptation to Pi starvation and is involved in negative feedback control of a subset of Pi marker genes. Our results reinforced the hypotheses that Pi-starvation-induced genes are under negative regulation, and the long-distance systemic signals controlling the Pi-starvation responses are dependent on whole-plant Pi status via negative modulation (Martin et al., 2000; Mukatira et al., 2001; Shin et al., 2006). However, links between AtSPX3 and Pi signaling have not been established.

Recently, bioinformatic approaches have identified SPX domains in transport homologs of the divalent anion:Na+ symporter family within the ion transporter superfamily (Barabote et al., 2006). As signal reception is one of the few non-transport functions that transport homologs have acquired over evolutionary time (Pinson et al., 2004), it is possible that most studied SPX proteins, such as SYG1 and PHO81, are involved in signal transduction. If it is supposed that the SPX domain functions as a sensor for Pi level in plants as it does for yeasts (Lenburg and O’Shea, 1996; Mouillon and Persson, 2006), the expression patterns and phenotypes shown here promote the hypothesis that AtSPX3 may have a role in sensing internal Pi deficiency, and AtSPX1 in sensing external Pi deficiency, both of which need to be further determined.

Figure 8 shows a scheme indicating the involvement of the AtSPX proteins in Pi-signaling pathways and their potential connections with some known components. PHR1, the MYB transcription factor, acts as a central factor that contributes to downstream Pi signaling by regulating the expression of a wide range of Pi-responsive genes (Rubio et al., 2001), including at least three of the four AtSPX genes. The transduction of Pi-starvation signals to PHR1 relies on the SUMO E3 ligase SIZ1 (Miura et al., 2005). AtSPX1 positively regulates the expression of the Pi-starvation-responsive genes ACP5, PAP2 and RNS1. AtSPX3, acting downstream of PHR1, may exert negative feedback regulation on uncharacterized factors upstream of PHR1. AtSPX3 may also exert negative regulation on AtSPX1 and six other genes, some of which are responsive to Pi starvation earlier than AtSPX3. The Pi-signaling pathways that AtSPX3 is involved in may also include microRNA regulation, as this vital factor for Pi homeostasis controls At4 expression (Bari et al., 2006; Chiou et al., 2006; Fujii et al., 2005; Shin et al., 2006).

Figure 8.

 A suggested model for AtSPX involvement in Pi-signaling pathways.
PHR1 is essential for inducing expression of AtSPX1–3 under Pi starvation. SIZ1, upstream of PHR1, is important for AtSPX1 and AtSPX2 expression (thick line), but not for AtSPX3 expression. AtSPX1 activates the expression of ACP5, PAP2 and RNS1 via interaction with uncharacterized transcription factor(s). AtSPX3, acting downstream of PHR1, exerts negative feedback regulation of AtSPX1 and other factor(s) upstream of PHR1. AtSPX3 repression results in upregulation of several Pi-responsive genes, including AtSPX1, RNS1, PAP2, IPS1, At4, AtPht1;4 and AtPht1;5. Arrowheads show positive regulation; flat-ended lines show negative regulation.

Previous studies have suggested that the SPX domain is most likely a domain for protein interaction (Kang and Ni, 2006), and the tomato IDS4-like protein interacts with a leucine zipper transcription factor (Sell and Hehl, 2005). Thus, the AtSPX family may perform their functions via direct protein binding for signal transduction. The molecular links described here may advance our understanding of the Pi-signaling pathway.

Experimental procedures

Plant growth

Wild type (Col-0) and transgenic seeds were surface-sterilized with 10% commercial bleach in 100% ethanol for 10 min, followed by several washes with 95% ethanol. All the plant media used here have been described previously (Ma et al., 2001). Agar B (0.7%) (Sangon Co. Ltd; was added to the media with a background Pi concentration of 1–2 μm. For expression profiling, seedlings germinated on normal medium (1 mm NH4H2PO4) were transferred once at 7 DAG to fresh medium, partially or completely replacing NH4H2PO4 with (NH4)2SO4. For phenotype analysis, a direct seeding strategy was adopted on media with varying Pi. Plates were sealed with parafilm and placed vertically in a growth cabinet (Percival Scientific;, with a 16 h light/8 h dark cycle, 150 μmol m−2 sec−1 photon flux density, and a constant temperature of 21.5°C.

Subcellular localization analysis

Full-length cDNA clones for the AtSPX genes were obtained from RIKEN BRC (pda07597 for AtSPX1, pda03041 for AtSPX4), the Arabidopsis Biological Resource Center (U68223 for AtSPX3) or directly amplified from Arabidopsis root cDNAs (for AtSPX2). pCambia1300 was modified in the our laboratory by introducing the CaMV 35S promoter between EcoRI and SacI, and a nos terminator between PstI and HindIII), and hereafter is referred as pCambia1300-mod. Full-length mgfp4 with additional SpeI and SalI restricted enzymes recognition sites was amplified from pBIN121 and digested with SpeI and SalI; full-length cDNAs before the stop codon of AtSPX1, AtSPX2 and AtSPX4 were amplified using the primer sets shown in Table S1 and digested with BamHI and SpeI. The purified GFP and gene fragments were synchronously introduced into pCambia1300-mod linearized with BamHI and SalI. The resultant constructs were confirmed by PCR and restriction analysis. For the AtSPX3–GFP fusion construct, the full-length gene before the stop codon was amplified (using the primers shown in Table S1) and digested with BglII. This fragment was then transferred to the BglII site of pCambia-1302. The resultant vector with correct insertion orientation was identified by restriction analysis.

All the GFP fusion constructs, together with original pCambia-1302, pBIN121 and pCAMB-RFPPTS1, were transiently expressed in onion epidermal cells using the Bio-Rad biolistic PDS-1000/He system (, performed essentially as described previously (Varagona et al., 1992). As for the transient expression assay in Arabidopsis cells, the mesophyll protoplasts were transformed by the polyethylene glycol (PEG) method (Sheen, 2001). For stable expression of GFP fusions, wild-type Col plants were transformed using the floral-dip method (Clough and Bent, 1998). The material used for GFP observation was the primary root of transgenic seedlings germinated on optimal medium for 5 days. The GFP fluorescence was imaged using a Carl Zeiss laser scanning system LSM510 (

Promoter–GUS constructs and histochemical analysis

The promoter 5′ UTR regions of the AtSPX genes were amplified from genomic DNA of Col using the primer pairs shown in Table S2. PCR products were first cloned into the pUCm-T vector (Shenerg Biocolor Co. Ltd; and sequenced for confirmation. Promoter fragments with correct insertion orientation were then transferred from pUCm-T to the expression vector pCambia1391Z between the cloning sites shown in Table S2. The resultant promoter–GUS constructs and the original pCambia1391Z were transformed into Col. Histochemical GUS staining was performed as described previously (Jefferson et al., 1987). To study the promoter activity in response to to Pi starvation, transgenic seeds were directly germinated on Pi-replete (1 mm) or Pi-deficient (15 μm) media, and analyzed at 1, 3, 5 and 7 DAG.

Over-expression of AtSPX1

The over-expression construct was generated by inserting a full-length AtSPX1 cDNA fragment into the binary vector pCambia1300-mod after the CaMV 35S promoter. The full-length AtSPX1 from the 5′ UTR to the 3′ UTR was amplified from cDNA clone pda07597 using primers aaaaggtaccCATTCTCACTTAAGTTTCCCAG and aaaatctagaTTACATCAACTACAACACACC (lowercase letters for site-directed cloning, not AtSPX1-specific nucleotides). The purified PCR product (1030 bp) was digested with KpnI and XbaI, and subsequently cloned into pCambia1300-mod linearized with KpnI and XbaI. Col plants were transformed using Agrobacterium tumefaciens EHA105, and transgenic plants were selected on hygromycin (20 mg l−1) medium. T3 progeny of two independent transgenic lines (T1) were used for phenotype study and expression analysis.

RNAi of AtSPX3

pBluescript was first tailored by introducing the second intron (215 bp) of the Phaseolus vulgaris nitrite reductase (NIR) gene between the SphI and PstI sites to generate the pBSin vector. A 440 bp AtSPX3 fragment was amplified using primers 5′-GTGGAATCTATTTTCGTCGGTT-3′ and 5′-CGCACATCTCTCGTATCCC-3′, and then cloned into pUCm-T. This fragment was twice transferred from pUCm-T into pBSin using the PstI and BamHI sites and the NsiI and XhoI sites sequentially, giving rise to two copies of the insertion with reverse orientations separated by an intron (hairpin-type structure). The intron-interrupted 600 bp hairpin fragment (including 112–309 bp of the AtSPX3 open reading frame, corresponding to the coding region for SPX subdomain II and the adjoining spacing sequences) released from pBSin by NcoI digestion was blunted and cloned into pCambia-1301 linearized using SmaI. Restriction analysis using EcoRI and HindIII was performed to confirm insertion of the hairpin loop into pCambia1301. The resultant expression vector was transformed into Col and more than 80 independent resistant T0 lines were generated. Transformants confirmed as AtSPX3-deficient were further propagated. T3 progeny of two independent transgenic lines (T1) were used for phenotype analysis.

RNA isolation, cDNA preparation and RT-PCR

RNA was extracted from shoots, roots or whole seedlings at various stages as indicated. First-strand cDNAs were synthesized from total RNA using SuperScript™ II reverse transcriptase (Invitrogen, Quantitative real-time PCR was performed using an ABI Prism 7000 sequence detection system (Applied Biosystems, Each 20 μl reaction contained 1 × SYBR Green master mix (Applied Biosystems), 0.5 μl cDNAs and 0.1 μl (20 μm) forward and reverse primers. The PCR conditions were as follows: 94°C for 2 min, followed by 40 cycles of 94°C for 15 sec, 58°C for 20 sec and 72°C for 30 sec. Fluorescence data were collected during the 72°C step and were analyzed using Sequence Detector version 1.7 software (Applied Biosystems). The amounts of cDNA template in each sample were normalized against that of actin. All primers for RT-PCR are shown in Table S3.

Measurement of P concentration in plants

Total phosphorus (P) was extracted from dried samples (30–50 mg) using sulfuric acid and hydrogen peroxide. Inorganic phosphate (Pi) in fresh samples (50–80 mg) was extracted using sulfuric acid. The P or Pi contents in plant sample solutions were analyzed 15 min after mixing with a malachite green reagent as described previously (Delhaize and Randall, 1995). The absorption values for the solutions at 650 nm were determined using a Spectroquant NOVA60 spectrophotometer (MERCK Co. Ltd.; Eight samples were used for each genotype, and two independent experiments were performed.


We thank the ABRC (Arabidopsis Biological Resource Center, Ohio State University, USA) for cDNA clone U68223 (AtSPX3) and T-DNA insertion mutants SALK_039445, 080503 and 019826 (AtSPX1, AtSPX2 and AtSPX4), and RIKEN BRC (RIKEN BioResource Center, Japan) for cDNA clones pda07597 (AtSPX1) and pda03041 (AtSPX4). Dr Javier Paz-Ares (Centro Nacional de Biotecnología, Campus de Cantoblanco, Spain) and Dr Kenji Miura (Center for Plant Environmental Stress Physiology, Purdue University, West Lafayette, IN, USA) kindly provided homozygous seeds for the phr1 and siz1 mutants, respectively. We also thank Dr Yun Lin of the University of Illinois (Department of Agriculture, Urbana-Champaign) for generous provision of the pCAMB-RFPPTS1 vector. The work was supported by the National Key Basic Research Program of China (2005CB120900) and National Natural Science Foundation of China (30500310).