SPX1 is an important component in the phosphorus signalling network of common bean regulating root growth and phosphorus homeostasis

Summary PvSPX1 was found to be a positive regulator in the P signalling network of common bean, and is downstream of PvPHR1.


Introduction
Phosphorus (P) is an essential element for plant growth, and is easily fixed by soil particles due to its chemical properties. Therefore, low P availability adversely affects crop growth and production, especially on acid soils (Raghothama, 1999;Vance et al., 2003). To cope with low P stress, plants have developed a wide range of adaptive strategies, such as changes in root architecture and morphology Zhou et al., 2008;Péret et al., 2011;Tian et al., 2012), increased exudation of protons and organic acids (Fox and Comerford, 1990;Ström et al., 2005;Taghipour and Jalali, 2012), and enhanced secreted or root-associated acid phosphatase activities (Del Pozo et al., 1999;Bozzo et al., 2002;Ligaba et al., 2004;C. Wang et al., 2009;Liang et al., 2010;Robinson et al., 2012;L.S. Wang et al., 2012). These adaptive strategies are tightly mediated by the P signalling network, which is composed of a wide array of regulators (Raghothama, 1999;Vance et al., 2003;Chiou and Lin, 2011).
In plants, four groups of proteins were also found to contain the SPX domain. Among them, three groups of proteins have the SPX domain in the N-terminus and other domains in the C-terminus, including an EXS (ERD1, XPR1, and SYG1), a major facility superfamily (MFS), or a RING-type zinc finger domain (Hamburger et al., 2002;Stefanovic et al., 2007;Lin et al., 2010;Secco et al., 2010;Kant et al., 2011;C. Wang et al., 2012). Similar to the functions of proteins containing the SPX domain in yeast, most of these plant members are involved in regulating P homeostasis in plants. Examples include OsSPX-MFS1 in rice (C. , along with AtPHO1:1 and AtNLA in Arabidopsis (Stefanovic et al., 2007;Secco et al., 2010;Kant et al., 2011).
Recently, a specific group of proteins only containing the SPX domain have been characterized in plants, such as Arabidopsis and rice (Duan et al., 2008;C. Wang et al., 2009;Z. Wang et al., 2009;Liu et al., 2010;Yang et al., 2011). In Arabidopsis, four members only contain the SPX domain, namely AtSPX1, AtSPX2, AtSPX3, and AtSPX4 (Duan et al., 2008). Furthermore, expression patterns of several Pi starvation-responsive genes were positively and negatively regulated by AtSPX1 and AtSPX3, respectively (Duan et al., 2008). Similarly, the negative regulatory role of OsSPX1 was also suggested in rice, because transcription of several Pi starvation-responsive genes (e.g. OsPT2, OsPT6, and OsPAP10) was suppressed through OsSPX1 overexpression (C. . Furthermore, it has recently been determined that OsSPX1 is downstream of OsPHR2 and OsPHO2 in the rice P signalling pathway (Liu et al., 2010).
Despite accumulated knowledge of the P signalling network in model plants (i.e. Arabidopsis and rice) (Chiou and Lin, 2011), information on Pi starvation-responsive pathways in other crops remains fragmentary. A group of Pi starvationresponsive genes (e.g. PvmiR399 and PvPS2:1) have been cloned and characterized in common bean (Phaseolus vulgaris L.), an important legume crop (Tian et al., 2007;Valdes-Lopez et al., 2008;Hernández et al., 2009;Liang et al., 2012a,b). This has facilitated elucidation of the P signalling network in bean, although this knowledge remains incomplete. Recently, essential roles for PvPHR1 and PvmiR399 have been suggested in P deficiency signalling (Valdes-Lopez et al., 2008). Nevertheless, other regulators are probably required as well. In a previous study, three expressed sequence tags (ESTs) with high homology to AtSPX1 were identified through screening a suppression subtractive hybridization library constructed from P-deficient bean (Tian et al., 2007). Among them, the full-length cDNA of PvIDS4-1 (i.e. PvSPX1) was cloned, and its expression levels were found to be up-regulated by Pi starvation in bean (Tian et al., 2007). However, the functions of PvSPX1 and other PvSPX genes in bean adaptation to P deficiency remain unknown. In this study, the full-length cDNA of the other two PvSPX genes (i.e. PvSPX2 and PvSPX3) was cloned. Subsequently, the expression patterns and functions of all three PvSPX gene family members as related to P availability were characterized in bean.

Materials and methods
Plant material and growth conditions Seeds of common bean genotype G19833 were surface sterilized for 1 min using 10% (v/v) H 2 O 2 and then germinated in the dark on germination paper moistened with 1/4 strength modified nutrient solution as described previously (Yan et al., 2001). Five days after germination, seedlings were transferred to nutrient solution supplied with 5, 50, 100, or 500 μM KH 2 PO 4 for a P dosage experiment. After 10 d, young leaves and roots were harvested. For time course experiments, seedlings were pre-treated in 1/4 strength nutrient solution for 7 d and then transplanted to nutrient solution containing 5 μM KH 2 PO 4 . Shoots and roots were each harvested at 0, 4, and 8 d after treatment for determination of fresh weight, total root length, and P content. Young leaves and roots were separately harvested for RNA extraction. Nutrient solution was well aerated and its pH was maintained between 5.8 and 6.0. Four biological replicates were included in all of the experiments.

Analysis of total root length and P content
Roots were scanned, and then the digital images were analysed using Win-Rhizo software (Régent Instruments, Canada) to measure total root length. Shoots and roots were kept separately at 75 °C until completely dry, and then were ground into powder for total P content analysis. P content was determined using the phosphorusmolybdate blue colour reaction as previously described (Murphy and Riley, 1962) Cloning full length cDNAs of PvSPX2 and PvSPX3 Gene-specific primers were designed according to the EST sequences of PvSPX2 (EG594307) and PvSPX3 (EG594308) (Tian et al., 2007) (Supplementary Table S1 available at JXB online). Using the full-length cDNA library constructed from the roots of G19833 as a template, the 5' and 3' termini of each gene were amplified by the specific primers paired with T3 and T7 primers, respectively. The amplified DNA fragments were then cloned into the pMD18-T vector (TaKaRa, Japan) and sequenced. Sequences of PvSPX2 and PvSPX3 were analysed at the National Center for Biotechnology Information (NCBI) website (http://www.ncbi.nlm.nih.gov/), and deposited in GenBank with accession numbers GU189405 for PvSPX2 and GU189406 for PvSPX3. Multiple sequence alignments were conducted using ClustalW 1.8. The phylogenetic tree was established using the Neighbor-Joining method of the MEGA 4.1 program.

RNA extraction and quantitative real-time PCR
Total RNA was isolated from young leaves and roots using RNAiso Plus reagent (TaKaRa) and treated with DNase I (TaKaRa). The first-strand cDNA was synthesized from total RNA using MMLV reverse transcriptase following the manual (Promega Inc., USA). The first-strand cDNA was then used for SYBR Green-monitored quantitative real-time PCR (qPCR) analysis, which was performed using a Rotor-Gene 3000 (Corbett Research, Australia). Expression levels of the tested genes were quantified relative to expression levels of the reference gene EF-1α (PvTC3216) using arbitrary units. The primer pairs used for qPCR analysis are shown in supplementary Table S1 at JXB online. All of the gene expression analyses had four biological replicates.

Subcellular localization analysis
The coding regions of PvSPX1 (EF191350), PvSPX2, and PvSPX3 without stop codons were separately cloned into the transient expression vector (pBEGFP), and fused with green fluorescent protein (GFP; Liang et al., 2010). For subcellular localization of PvSPXs in onion (Allium cepa) epidermal cells, the PvSPX-GFP fusion constructs and GFP empty vector control were separately transformed into onion epidermal cells using a helium-driven accelerator (PDS/1000, Bio-Rad). After the transformed cells were cultured on Murashige and Skoog (MS) medium for 16 h, the GFP florescence was observed using a confocal scanning microscope system (TCS SP2, Leica, Germany) with 488 nm excitation and 500-525 nm emission filter wavelengths. For subcellular localization of PvSPXs in leaf epidermal cells of tobacco (Nicotiana tabacum), the PvSPX-GFP fusion constructs and GFP empty vector control were separately transformed into Agrobacterium tumefaciens strain GV3101, which were further used for transformation as previously described (Sparkes et al., 2006). After the transformation, plants were grown under normal conditions for 48 h and the GFP florescence was observed using a fluorescence microscope (Leica DM5000B). The GFP fluorescence was imaged using a Leica DFC 480 camera.

Transformed genes in common bean hairy roots
The coding regions of PvSPX1, PvSPX2, PvSPX3, and PvPHR1 (EU500763) were inserted separately into the unique BamHΙ and MluΙ sites of the binary vector pYLRNAi as previously described (Liang et al., 2010). For PvSPX1 RNA interference (RNAi) construction, the same binary vector pYLRNAi was used by inserting the PvPSX1-specific fragment into the BamHΙ and HindIII, and the PstI and MluI sites, respectively. The overexpression, RNAi constructs, and the empty vector control (CK) were then separately transformed into Agrobacterium rhizogenes strain K599, which were further used for hairy root transformation. Transformed bean hairy roots were generated and maintained as described previously (Liang et al., 2012b). Briefly, sterilized bean seeds were germinated on halfstrength MS medium. After 35 h, the abaxial sides of cotyledons were wounded with a scalpel previously dipped into the overnight cultures of the transgenic A. rhizogenes strain K599. The wounded cotyledons were cultured in solid MS medium to develop hairy roots. The expression levels of the corresponding genes in hairy roots were verified through qPCR analysis. For P treatments, ~0.2 g (fresh weight) of hairy roots was cultured in solid MS medium with or without the addition of 1.25 mM KH 2 PO 4 . After 14 d growth, transgenic bean hairy roots were photographed using a microscope (Leica) and a Leica DFC 480 camera. The fresh weight and total P content of each transgenic line were determined as described above. The lateral root length was analysed using Win-Rhizo. Based on root hair density, lateral roots were separated into two zones, namely the root hair zone (i.e. the part of the root zone with >10 root hairs per 1 mm root) and the non-root hair zone, and then the percentage of the root hair zones in the lateral roots was calculated. In total, 10 lateral roots were analysed for each replicate. For each treatment, four biological replicates were included.
To analyse the expression patterns of genes downstream of PvSPX1 in P signalling, total RNA was extracted from transgenic hairy roots grown under high P conditions. Subsequently, qPCR was conducted to analyse the expression of 11 genes downstream of PvLPR1-like (FE710903), and PvPDR2-like (TC44308). All qPCR primers (Supplementary Table S1 at JXB online) were designed according to the sequences downloaded from the Dana-Farber Cancer Institute (DFCI; http://compbio.dfci.harvard.edu/ tgi/) for PvPHT2, PvLPR1-like, and PvPDR2-like, or from the GenBank database (http://www.ncbi.nlm.nih.gov/genbank) entries for PvPHR1, PvPT1,Pv4,PvPAP1,PvPAP2,PvPAP3,PvPAP4,PvPAP5,and PvPS2:1. To construct the PvSPX1 and PvSPX2 promoter fused with β-glucuronidase (GUS) vectors, the 5'-regulatory regions of 1.9 kb for PvSPX1 and 1.6 kb for PvSPX2 were each cloned, and inserted into the pCAMBIA 1391 vector. The constructs were transformed into the bean hairy roots as described above, and GUS activity was analysed as described before (Liang et al., 2012a).

Results
Plant growth was affected by P availability P deficiency significantly inhibited bean growth, as reflected by decreases in plant fresh weight, total root length, and total P content (Supplementary Table S2 at JXB online). With an increased duration of P deficiency, plant fresh weight and total P content gradually decreased. After 8 d of Pi starvation, the total P contents of bean shoots and roots were reduced by 32% and 30%, respectively, as compared with high P conditions (Supplementary Table S2). Similarly, total root length was also significantly decreased by P deficiency. Total root length at 4 d and 8 d of P deficiency was reduced by 39% and 55%, respectively, as compared with under high P conditions (Supplementary Table S2).

Identification and bioinformatics analysis of PvSPX2 and PvSPX3
Based on the reported EST sequences of PvSPX2 and PvSPX3, the full-length cDNAs of both PvSPX2 and PvSPX3 were cloned from a full-length cDNA library of G19833 subjected to P deficiency. The coding regions of PvSPX2 and PvSPX3 were 861 bp and 756 bp in length, respectively. Alignment analysis showed that PvSPX2 and PvSPX3 exhibited 75% and 50% similarity to PvSPX1, respectively.
Phylogenetic analysis showed that plant proteins containing the SPX domain could be divided into four groups, namely SPX, SPX-EXS, SPX-MFS, and SPX-RING (Fig. 1). Furthermore, SPX proteins could be further subdivided into three groups. Among them, PvSPX1, PvSPX2, and PvSPX3 belong to group I, which includes AtSPX1 and AtSPX2 in Arabidopsis, as well as OsSPX1 and OsSPX2 in rice (Fig. 1).

Temporal expression patterns of PvSPX genes in response to Pi starvation
The temporal expression patterns of the three PvSPX genes in bean leaves and roots were analysed by qPCR. As shown in Fig. 2, their expression levels were significantly increased over time and reached their highest levels after 8 d of low P treatment (Fig. 2). However, their expression patterns varied in leaves and roots at 4 d of P deficiency (Fig. 2). After 4 d of P deficiency, significantly increased transcription was observed for PvSPX1 and PvSPX2 in leaves, while for PvSPX3 transcription was not increased either in leaves or in roots (Fig. 2). This suggests that PvSPX1 and PvSPX2 respond to Pi starvation earlier than PvSPX3 in bean.

Dosage responses of PvSPX genes to P availability
Expression patterns of the three PvSPX members studied here were tightly dependent on P availability in the medium (Fig. 3). Their highest transcript levels were observed in both leaves and roots supplied with 5 μM P, and were decreased with increased P availability (Fig. 3). When the applied P concentration was increased to 500 μM, transcription of each PvSPX gene was negligible (Fig. 3). However, slight differences existed among their expression patterns as related to P availability. Transcript levels of PvSPX1 and PvSPX2 in both leaves and roots declined significantly when the applied P concentration was increased from 100 μM to 500 μM, but that of PvSPX3 did not (Fig. 3), suggesting that expression of PvSPX1 and PvSPX2 might be more sensitive to P availability than that of PvSPX3.

Subcellular localization of PvSPX proteins
To determine the subcellular localization, the coding regions of the three PvSPX genes were fused with the GFP reporter gene and transiently expressed in onion and tobacco epidermal cells. Subcellular localization was visualized by detecting GFP signal in the transformed onion and tobacco epidermal cells. The empty vector containing 35S:GFP was used as a control. The results showed that the three PvSPX members were found in various subcellular localizations (Fig. 4). Signals of GFP fusion with PvSPX1 and PvSPX2 were only detected in the nuclei of onion and tobacco epidermal cells (Fig. 4). However, GFP fusion with PvSPX3 was observed in many areas in onion and tobacco epidermal cells, suggesting  that PvSPX3 might be localized in the cytoplasm and nuclei (Fig. 4).

Functional analysis of PvSPX genes in transgenic hairy roots
The functions of PvSPX genes were further analysed in bean transgenic hairy roots by overexpressing PvSPX1, PvSPX2, and PvSPX3. Significantly increased transcripts of the three PvSPX genes in the transgenic bean hairy roots were verified through qPCR analysis ( Supplementary Fig.  S1 at JXB online). Subsequently, the transgenic hairy roots were grown in MS medium with or without P application for 14 d. The results showed that only overexpressing PvSPX1 could inhibit hairy root growth, as reflected by reduced fresh weight of hairy roots under both P conditions (Fig. 5A, B). Compared with the control lines, the fresh weight of the PvSPX1 overexpression line was reduced by ~60% in high P and 40% in low P (Fig. 5B). Furthermore, the P concentration in the PvSPX1 overexpression line was higher than that in the control line by ~45% in high P and 30% in low P (Fig. 5C). In contrast, the fresh weight and P concentration of both PvSPX2 and PvSPX3 overexpression lines were similar to those in the control line at the two P levels (Fig. 5B, C). Similarly, suppressed PvSPX1 did not affect hairy root fresh weight and P concentration, compared with those in the empty vector (CK) controls (Supplementary Figs S2, S3). Root morphology was further investigated in all hairy root lines at the two P levels through determination of the percentage of the root hair zones (i.e. the part of root zone with >10 root hairs per 1 mm root) in bean hairy roots. The percentage of the root hair zones in all hairy root lines was ~80% without P application (Fig. 6). With P application, the percentage of the root hair zones of CK, and PvSPX2 and PvSPX3 overexpression lines was decreased by >50% (Fig. 6). However, for the PvSPX1 overexpression line, applied P did not affect the percentage of the root hair zone (Fig. 6). Interestingly, similar results were also observed in PvSPX1 RNAi lines, in which the percentage of the root hair zone was not affected by P application ( Supplementary  Fig. S4 at JXB online). The results suggest that expression of PvSPX1 might regulate enlargement of the root hair zones at a high P level.

PvSPX participates in the P signalling network in bean
The expression patterns of 11 genes were investigated in the transgenic hairy roots overexpressing PvSPX1 in order to illustrate the regulatory role of PvSPX1 in the P signalling  Fig. 3. Dosage response of PvSPX genes to P deficiency. Seedlings were grown in nutrient solution supplied with 5, 50, 100, or 500 μM KH 2 PO 4 . After 10 d, total RNA was isolated from leaves and roots for qPCR analysis. Expression levels of the tested genes were quantified relative to expression levels of the reference gene EF-1α (PvTC3216) using arbitrary units. Each bar is the mean of four replicates with the standard error. Different letters represent significant differences at the 0.05 level.
network in bean. Among them, nine genes were previously characterized as Pi starvation-responsive genes, namely two Pi transporters (PvPT1 and PvPHT2), five purple acid phosphatases (PvPAP1-PvPAP5), Pv4, and PvPS2:1. The other two genes (PvLPR1-like and PvPDR2-like) exhibit high homology with AtLPR1 and AtPDR2, respectively, which both regulate root growth in Arabidopsis. The qPCR analysis showed that overexpressing PvSPX1 led to significantly increased transcription of 10 genes compared with the control line, namely PvPT1, PvPHT2, Pv4, PvPAP1-PvPAP5, PvPS2:1, and PvLPR1-like (Fig. 7). Consistently, suppressed transcripts of PvSPX1 resulted in lower expression patterns of several genes-PvPHT2, PvPAP3, PvPS2:1, and PvLPR1like ( Supplementary Fig. S5 at JXB online). The results suggest that expression of these genes is positively regulated by PvSPX1. However, expression levels of PvPDR2-like were inhibited in the PvSPX1 overexpression lines and increased in the PvSPX1 RNAi lines, compared with those in the control line ( Fig. 7; Supplementary Fig. S7), suggesting that PvPDR2-like is negatively regulated by PvSPX1 in bean. Similarly, PvSPX2 overexpression resulted in increased transcripts of several genes downstream of PvSPX1, except PvPDR2-like ( Supplementary Fig. S6), suggesting that PvSPX2 might have a similar regulatory role to PvSPX1. However, overexpression of PvSPX3 did not affect expression patterns of genes downstream of PvSPX1 ( Supplementary  Fig. S6). Furthermore, significantly increased transcription of PvSPX1 was obviously observed in the transgenic bean hairy roots with overexpression of PvPHR1 (Fig. 8), suggesting that PvSPX1 lies downstream of PvPHR1. Fresh weight of bean hairy roots at two P levels. (C) P concentration in bean hairy roots. Bean hairy roots were grown in media containing 0 μM (-P) or 1.25 mM (+P) KH 2 PO 4 for 14 d. THe fresh weight and P concentration were measured. Each bar is the mean of four replicates with the standard error. Asterisks represent significant differences between overexpressing PvSPX and CK for the same trait in t-tests. *0.01<P≤0.05; **P≤0.01. OX1 and OX2 indicate two transgenic bean hairy root lines overexpressing PvSPX1, PvSPX2, or PvSPX3. CK1 and CK2 indicate the two transgenic lines transformed with the empty vector.

Discussion
Proteins containing the SPX domain have been well documented to be involved in the P signalling pathway of yeast and model plants, including Arabidopsis and rice (Lenburg and O'Shea, 1996;Duan et al., 2008;C. Wang et al., 2009;Z. Wang et al., 2009;Lin et al., 2010;Secco et al., 2012a). However, involvement of SPX proteins in P signalling remains largely unknown in legumes. In this study, three PvSPX genes were cloned and comparatively characterized as related to Pi starvation in bean. The results demonstrated that PvSPX1 is an important regulator in the P signalling network of common bean, which shows several novel functions in regulating root growth, P homeostasis, and downstream gene transcription. Since the transcription of several Pi starvation-responsive genes was noticeably increased and decreased in the PvSPX1 overexpression and RNAi transgenic bean hairy roots, respectively, PvSPX1 appears to be a positive regulator in the bean P signalling network ( Fig. 7; Supplementary Fig. S7 at JXB online). Furthermore, PvSPX1 appears to be a downstream gene of PvPHR1, because overexpressing PvPHR1 led to increased transcription of PvSPX1 in bean hairy roots (Fig. 8). Similarly, it has been demonstrated that AtSPX1 and OsSPX1 were downstream genes of AtPHR1 and OsPHR2 in the P signalling pathways of Arabidopsis and rice, respectively (Duan et al., 2008;C. Wang et al., 2009;Liu et al., 2010). However, regulatory roles of PvSPX1, AtSPX1, and OsSPX1 in the P signalling pathways seemed to vary among species despite them showing several similar properties, such as nuclear localization and Pi starvation-induced expression patterns (Figs 2-4). In rice, OsSPX1 has been considered as a negative regulator in the P signalling network because overexpressing OsSPX1 significantly suppressed the expression levels of 10 Pi starvation-induced genes (C. . Also, OsSPX1 suppression resulted in increased transcripts of OsPT2 and OsPT8 in rice (C. Liu et al., 2010). However, AtSPX1 was considered as a positive regulator in the Arabidopsis P signalling pathway because overexpressing AtSPX1 led to increased transcription of several genes increased by Pi starvation, such as AtACP5 and AtRNS1 (Duan et al., 2008). Therefore, it seems that the regulatory roles of SPX1 in dicots might differ from those in monocot plants, which needs to be further studied.
Consistent with the enhanced expression levels of two Pi transporter genes (PvPHT2 and PvPT1), a significantly increased P concentration was observed in bean hairy roots overexpressing PvSPX1, especially under high P conditions (Fig. 5C). This suggests that PvSPX1 is involved in regulating P homeostasis in bean roots. Similarly, it has been documented that suppressed expression of OsSPX1 led to more P accumulation in both leaves and roots in rice under high P conditions (C. Liu et al., 2010). Taken together, PvSPX2 PvSPX3 Fig. 6. Root morphology and the percentage of the root hair zones in bean hairy roots at two P levels. (A) Photographs of roots in bean hairy roots at two P levels. (B) Percentage of the root hair zone in bean hairy roots at two P levels. Bean hairy roots were grown in media containing 0 μM (-P) or 1.25 mM (+P) KH 2 PO 4 for 14 d. Ten lateral roots were selected from each replicate for further analysis. Photographs were taken of three zones of the lateral roots, namely the root tip, root middle, and root base. Scale bars=1 mm. OX1 and OX2 indicate two transgenic bean hairy root lines overexpressing PvSPX1, PvSPX2, or PvSPX3. CK1 and CK2 indicate the two transgenic lines transformed with the empty vector. Asterisks represent significant differences between two P treatments for the same trait at the 0.05 level.
Relative expression these results suggest that SPX might control P homeostasis in plants through regulating expression of Pi transporter (PT) genes. However, the molecular mechanisms underlying SPX regulation of PT transcription remain largely unknown. Since PvSPX1 has regulatory roles which appear to contrast with those of OsSPX1 and AtSPX3 in P signalling pathways, it is plausible that SPX might not directly control downstream gene expression. It will be important to clarify the functions of SPX through identification of other P signalling regulators interacting with SPX in plants.
Another novel feature of PvSPX1 is its involvement in regulating root growth and root morphology in bean roots. Changes in root morphology, such as inhibition of root elongation and stimulation of root hair growth, are well accepted as typical responses of plant roots to Pi starvation (Péret et al., 2011). It was found here that overexpressing PvSPX1 significantly inhibited hairy root growth at two P levels (Fig. 5B), but led to enlarged root hair zones (Fig. 6). This suggests that overexpression of PvSPX1 could enhance root morphological modifications in adaptation to P deficiency.
In bean, it has been documented that the P-efficient genotype G19833 has greater root hair density and longer root hair length than the P-inefficient genotype DOR364 in low P conditions . In this study, the Pi starvation-induced PvSPX1 was originally cloned from G19833. Furthermore, higher PvSPX1 expression levels were found in G19833 than in DOR364 (data not shown) at low P, suggesting positive contributions of PvSPX1 to superior P efficiency in G19833 through regulation of root morphology. Subsequently, two genes regulating root growth in transgenic bean hairy roots, PvLPR1-like and PvPDR2like, were cloned and their transcription was investigated Overexpression of PvSPX1 led to increased expression of PvLPR1-like and reduced expression of PvPDR2-like (Fig. 7). Since it has been documented that AtLPR1 and AtPDR2 are two critical components regulating root growth in opposite ways (Ticconi et al., 2004(Ticconi et al., , 2009Reymond et al., 2006;Svistoonoff et al., 2007;Wang et al., 2010;Miura et al., 2011), it is conceivable that changes in root morphology in bean result from up-regulation of PvSPX1, with consequent effects on transcripts of PvLPR1-like and PvPDR2-like.
Although the three PvSPX proteins studied here exhibit high homology, and belong to the same subgroup in phylogenetic tree analysis (Fig. 1), diverse properties and functions of PvSPXs were observed in response to Pi starvation, as reflected by different expression patterns, variations in subcellular localization, and dissimilar growth of transgenic bean hairy roots. In response to Pi starvation, it seems that PvSPX1 and PvSPX2 might be earlier responsive genes which are more sensitive to Pi starvation than PvSPX3 in bean leaves. At 4 d of Pi starvation, transcripts of both  PvSPX1 and PvSPX2 in bean leaves were significantly increased, while PvSPX3 remained unchanged (Fig. 2). Also, with an increase in available Pi from 100 μM to 500 μM in the medium, significantly decreased transcription was observed for PvSPX1 and PvSPX2, but not for PvSPX3 (Fig. 3). Interestingly, through their promoter-fused GUS activity analysis in bean hairy roots, it was found that PvSPX1 and PvSPX2 exhibited similar spatial expression patterns ( Supplementary Fig. S7 at JXB online), and similar responses to P deficiency as well as subcellular localization (Figs 2, 4), but only overexpression of PvSPX1 resulted in inhibited root growth, increased root P concentration, and changes of morphological traits in transgenic bean hairy roots (Figs 5, 6), strongly suggesting diverse functions of PvSPX members, and PvSPX requirement of other P signalling regulators to regulate P homeostasis and root growth in bean.
Similarly, diverse functions of SPX members have been demonstrated in Arabidopsis and rice (Duan et al., 2008;C. Wang et al., 2009). In Arabidopsis, suppressed AtSPX3 led to an increased P concentration in shoots, and aggregative responses to Pi starvation (Duan et al., 2008). However, knock-down of AtSPX1, AtSPX2, or AtSPX4 did not alter the phenotypes of Arabidopsis at two P levels (Duan et al., 2008). In rice, suppressed plant growth was observed through overexpression of OsSPX1 and OsSPX3, as well as suppression of OsSPX1 (C. Z. Wang et al., 2009). However, functions of other SPX members as related to P deficiency in rice still remain unknown.
Taken together, the results demonstrate that PvSPX1 is a positive regulator in the P signalling network of common bean, and is downstream of PvPHR1 (Fig. 9). Increased transcription of PvSPX1 led to significantly coordinated expressions of a group of Pi starvation-responsive genes, which could dramatically regulate changes of root morphology, Pi acquisition and mobilization, as well as P homeostasis in bean roots (Fig. 9).

Supplementary data
Supplementary data are available at JXB online. Figure S1. Expression of PvSPX genes in transgenic bean hairy roots. Figure S2. Expression of PvSPX1 in PvSPX1 RNAi transgenic bean hairy roots. Figure S3. Growth and P concentration of bean hairy roots in CK and PvSPX1 RNAi transgenic lines at two P levels. Figure S4. Percentage of root hair zone in bean hairy roots with suppressed PvSPX1 at two P levels. Figure S5. Transcription levels of downstream genes of PvSPX1 in CK and PvSPX1 RNAi transgenic lines. Figure S6. Transcription levels of downstream genes of PvSPX1 in CK and overexpression transgenic lines of PvSPX2 or PvSPX3. Figure S7. Expression patterns of PvSPX1 and PvSPX2 through their promoter:GUS analysis. Table S1. List of primers used in the study.