Arabidopsis Pht1;5 Mobilizes Phosphate between Source and Sink Organs and Influences the Interaction between Phosphate Homeostasis and Ethylene Signaling1[W][OA]

Phosphorus (P) remobilization in plants is required for continuous growth and development. The Arabidopsis (Arabidopsis thaliana) inorganic phosphate (Pi) transporter Pht1;5 has been implicated in mobilizing stored Pi out of older leaves. In this study, we used a reverse genetics approach to study the role of Pht1;5 in Pi homeostasis. Under low-Pi conditions, Pht1;5 loss of function (pht1;5-1) resulted in reduced P allocation to shoots and elevated transcript levels for several Pi starvation-response genes. Under Pi-replete conditions, pht1;5-1 had higher shoot P content compared with the wild type but had reduced P content in roots. Constitutive overexpression of Pht1;5 had the opposite effect on P distribution: namely, lower P levels in shoots compared with the wild type but higher P content in roots. Pht1;5 overexpression also resulted in altered Pi remobilization, as evidenced by a greater than 2-fold increase in the accumulation of Pi in siliques, premature senescence, and an increase in transcript levels of genes involved in Pi scavenging. Furthermore, Pht1;5 overexpressors exhibited increased root hair formation and reduced primary root growth that could be rescued by the application of silver nitrate (ethylene perception inhibitor) or aminoethoxyvinylglycine (ethylene biosynthesis inhibitor), respectively. Together, these data indicate that Pht1;5 plays a critical role in mobilizing Pi from P source to sink organs in accordance with developmental cues and P status. The study also provides evidence for a link between Pi and ethylene signaling pathways.


INTRODUCTION 1
Arabidopsis (Himelblau and Amasino, 2001). Therefore, the translocation of Pi into sink 1 tissues/cells is important for sustaining growth under low Pi conditions. Chloroplast localized 2 Pht2;1, a low-affinity Pi transporter (Km of ~ 0.4 mM), has been shown to mediate Pi 3 translocation within the aerial parts of Arabidopsis (Daram et al., 1999;Versaw and Harrison, 4 2002). Low-affinity transporters from barley (HvPht1;6) and rice (OsPht1;2) have also been 5 implicated in Pi remobilization from leaves and Pi movement from root to shoot, respectively 6 (Rae et al., 2003;Ai et al., 2009;Preuss et al., 2010). Despite these developments, little is known 7 about the molecular mechanisms that govern Pi translocation and remobilization in higher plants. 8 Spatial expression patterns of Pht1 members in different tissue types and organs of 9 Arabidopsis suggest their potential involvement not only in Pi acquisition, but also in internal Pi 10 distribution to metabolically active and growing parts of the plant (Karthikeyan et al., 2002(Karthikeyan et al., , 11 2009Mudge et al., 2002;Miller et al., 2009). Among the Pht1 members, Pht1;5 showed Pi-12 deficiency-induced expression specifically in the phloem cells of older leaves, cotyledons, and 13 flowers (Mudge et al., 2002). Genome-wide transcriptome analysis further corroborated the 14 expression of Pht1;5 over the course of developmentally-regulated senescence in the leaves of 15 Arabidopsis (van der Graaf et al., 2006). However, the functional characterization of Pht1;5 and 16 its potential role in Pi translocation/remobilization has not been elucidated. Here, we used loss-17 of-function mutants of Pht1;5 and transgenic lines overexpressing this gene in Arabidopsis to 18 demonstrate its role in Pi mobilization between source and sink under different Pi regimes. We 19 also provide evidence for a tangible link between Pi transporters and ethylene signaling. 20

Expression profile of Pht1;5, and isolation of pht1;5 T-DNA insertion mutants 23
Previous studies have demonstrated the Pi-starvation-responsiveness of Pht1;5 expression in 24 Arabidopsis (Mudge et al., 2002;Morcuende et al., 2007;Thibaud et al., 2010). To further 25 determine the spatial expression pattern of Pht1;5, wild-type (WT) Arabidopsis plants were 26 germinated hydroponically on one-half-strength MS medium for 5 d and transferred to low (10 27 µM Pi, P-) and high Pi (1250 µM Pi, P+) for 7 d. Quantitative reverse-transcription PCR (qRT-28 PCR) analysis was used to measure Pht1;5 transcript levels in different tissues (roots and rosette 29 leaves) grown under P+ and P-conditions. As shown in Figure 1A, Pi deficiency triggered 30 increases in the abundance of Pht1;5 transcripts in both roots and rosette leaves. Consistent with 31 earlier studies (Karthikeyan et al., 2002;Mudge et al., 2002;Shin et al., 2004), Pht1;4 transcript 1 levels increased strongly in response to Pi deficiency in roots and rosette leaves (Fig. 1A). The 2 data thus validated the fidelity of the growth condition used for elucidating the effect of Pi 3 deficiency on the spatial expression profile of Pht1;5. 4 To determine the role of Pht1;5 in acquisition and mobilization of Pi, a reverse genetics 5 approach was employed. Two homozygous mutants [i.e., SALK_074836 (pht1;5-1) and 6 SALK_138009C (pht1;5-2)] were identified with T-DNAs inserted at 919 bp (exon) and 1020 bp 7 (intron), respectively, downstream of the translation start site of the Pht1;5 gene ( Fig. 1B). 8 Genetic analysis revealed that the pht1;5-1 T-DNA segregated as a single insertion locus (data 9 not shown), whereas estimation of the T-DNA copy number of pht1;5-2 (a confirmed 10 homozygous insertion line, ABRC) via Quantitative PCR revealed the presence of ~5 and ~7 11 copies of nptII and T-DNA left border compared to pht1;5-1 (Supplemental Table S2).  was performed on seven-day-old Pi-starved WT, pht1;5-1, and pht1;5-2 seedlings for 13 determining the levels of Pht1;5 expression ( Fig. 1C). An amplified product corresponding to 14 Pht1;5 was detected in WT, whereas no amplification products were detected in pht1;5-1 or 15 pht1;5-2. Further, qRT-PCR analyses of Pi-starved roots and shoots found no detectable 16 amplification for Pht1;5 in these mutants (data not shown). The results confirmed the 17 identification of two independent loss-of-function mutants for Pht1;5. 18 19

Loss-of-function mutation of Pht1;5 results in altered Pi allocation between shoot and root 20
During Pi deficiency, the levels of Pht1;5 transcripts increased appreciably in both roots and 21 rosette leaves (Fig. 1A). Therefore, a role for Pht1;5 in translocating Pi between root and shoot 22 was hypothesized. To test this, radiolabeled 33 Pi was used to compare the rates of root-to-shoot 23 Pi translocation among WT, pht1;5-1, and pht1;5-2. A loss-of-function mutant (pht1;1-2) of 24 high-affinity transporter Pht1;1, which affects both Pi acquisition and allocation (Shin et al., 25 2004), was included for comparison. Under P+ conditions, the distribution of 33 Pi in the shoots 26 was comparable among WT, pht1;5-2, and pht1;1-2, and was marginally higher for pht1;5-1 27 ( Fig. 2A). However, under P-conditions, both pht1;5 mutants showed 35-40% lower 33 Pi 28 distribution to the shoots relative to WT. Although 33 Pi distribution in pht1;1-2 was also 29 significantly (P<0.05) lower than WT, it was significantly (P<0.05) higher than the pht1;5 30 mutants ( Fig. 2A). These results suggest a more pronounced role for Pht1;5 in allocating Pi to the 31 shoots during Pi deprivation. To determine whether the attenuated shoot Pi translocation rates 1 observed in the mutants correlated with a change in P accumulation, we compared the total shoot 2 P content of WT, pht1;5-1, and pht1;1-2 (Fig. 2B). Pi deficiency resulted in significant 3 reductions (P<0.05) in the total shoot P accumulation in all genotypes. Although the P content in 4 the P+ shoots of pht1;5-1 was higher than in WT, under P-conditions both mutants (pht1;1-2 5 and pht1;5-1) accumulated about 20% less P in their shoots compared to WT (Fig 2B). Shoot P 6 measurements in leaves and floral stalks from Pi-starved pht1;5-1 plants show similar decreases 7 compared to WT when grown hydroponically under greenhouse conditions (Supplemental Fig.  8   S1). The decrease in shoot P content in pht1;5-1 was amended by complementing the mutant 9 with Pht1;5 cDNA (Supplemental Fig. S1). The data support a role for Pht1;5 in facilitating 10 mobilization of Pi between root and shoot independently, or in conjunction with, Pht1;1. 11 Whether altered Pi mobilization in the shoots of the pht1;5 mutant has a commensurate effect on 12 Pi-starvation-response (PSR) gene expression was determined via qRT-PCR analysis of several 13 PSR genes (Misson et al., 2005;Shin et al., 2006). Under P+ conditions, other than a marginal 14 increase in At4 transcripts in the pht1;5-1 mutant, the relative transcript levels for the PSR genes 15 tested (At4, Pht1;4, DGD1, and SQD1) were similar between pht1;5-1 and WT (Fig. 2C). In 16 contrast, the transcript levels for all the genes tested were significantly higher in the Pi-deprived 17 shoots of pht1;5-1 compared to WT. These data corroborate that a more aggravated Pi starvation 18 response is being experienced by the shoots of pht1;5-1 compared to WT due to the disruption of 19 Pi mobilization from root to the shoot during Pi deficiency. 20 To gain further insight into a possible role for Pht1;5 in the maintenance of Pi 21 homeostasis, 33 Pi uptake rate in the roots was evaluated for WT,pht1;pht1;and pht1;[1][2] seedlings grown under P+ and P-conditions (Fig. 3A). Independent of the Pi regime, there were 23 significant (P<0.05) increases (~10-20%) in the 33 Pi uptake rates (i.e. net accumulation of 33 Pi) 24 in the roots of pht1;5-1 and pht1;5-2 compared to that of WT. On the contrary, there was a 25 significant (P<0.05) decline (~30%) in root Pi uptake for pht1;1-2 compared to WT when grown 26 under Pi-replete conditions, which is consistent with an earlier study (Shin et al., 2004). Despite 27 differential 33 Pi uptake rates observed for the pht1;1 and pht1;5 mutants under variable Pi 28 conditions, the total P content in the roots of these mutants under both P+ and P-conditions were 29 very similar and significantly (P<0.05) lower compared to WT (Fig. 3B). The total P content 30 data for roots (Fig. 3B) and shoots (Fig. 2B) of WT, pht1;5-1, and pht1;1-2 were used to 31 calculate root:shoot distribution ratios of P (Fig. 3C). Although Pi deficiency exerted no 1 significant (P<0.05) differences on the ratio among WT and the mutants, pht1;5-1 showed a 2 significant (P<0.05) decline in the ratio compared to WT and pht1;1-2 when grown under Pi-3 replete conditions. Considering that Pht1;5 is expressed strongly in shoot tissues during Pi-4 replete conditions (Fig 1A), these data suggest that Pht1;5 is involved in mobilization of Pi from 5 shoots to roots during high Pi conditions. 6 Arsenate (AsO 4 3-) is an oxyanion structurally analogous to Pi and is taken up by roots via 7 high-affinity Pi transporters (Asher and Reay, 1979;Shin et al., 2004;Catarecha et al., 2007). To 8 gain further insight into the role of Pht1;5 in Pi acquisition, we compared the phenotypic 9 responses of the pht1;1-2 and pht1;5-1 mutants to arsenate (Fig. 4). Although the toxic effect of 10 arsenatewas evident on the growth of both the mutants and WT, the degree of tolerance varied. 11 The shoot fresh weight of WT grown on arsenate was 42% compared to that of untreated WT 12 seedlings (Fig. 4A). The corresponding values for the pht1;1-2 and pht1;5-1 mutants were 77% 13 and 65%, respectively, which indicated greater tolerance to arsenate by the mutants relative to 14 WT. A similar trend of higher tolerance in the mutants was evident with respect to lateral root 15 development (Fig. 4B). Together these data indicate that loss of Pht1;5 confers weak tolerance to 16 arsenate, suggesting that Pht1;5 could influence acquisition of both Pi and arsenate. To further examine the in planta role of Pht1;5 in Pi allocation we overexpressed the 21 Pht1;5 coding region under the control of the ACTIN 2 (ACT2) promoter in WT Arabidopsis. 22 The ACT2 promoter imparts strong constitutive expression in vegetative tissues (An et al., 1996;23 Kandasamy et al., 2002). Two independently generated transgenic lines (5A and 11C) with 24 relatively high levels of Pht1;5 expression compared to WT (Supplemental Fig. S2) were 25 selected for phenotypic characterization. After four weeks of growth the Pht1;5-overexpressors 26 showed substantial increases in shoot biomass and leaf area compared to WT and pht1;5-1 (Fig.  27 5A). A more detailed phenotypic characterization was carried out on WT, pht1;5-1, and 28 overexpressors that were grown for an additional two weeks. The Pht1;5-overexpressors 29 exhibited significant (P<0.05) increases in total leaf area (Fig. 5B), floral stalk thickness (Fig.  30 5C), and total leaf dry weight (Fig. 5D) compared to WT and pht1;5-1. On the other hand, 31 pht1;5-1 initially displayed slower growth at the four week-stage ( Fig. 5A) but morphological 1 parameters after two more weeks were only marginally affected compared to the WT (Fig. 5, B  2 and D). These phenotypic differences were observed under both short- (Fig. 5) and long-day (not 3 shown) growth conditions. 4 To examine the effect of Pht1;5-overexpression on P distribution and acquisition, the 5 total shoot and root P content and 33 Pi root uptake rates of WT and Pht1;5-overexpressors were 6 determined. Seedlings were grown on agar-solidified Pi-replete medium. Relative to WT, there 7 was a significant (P<0.05) decline in the total P content in the shoots of both overexpressors, but 8 an increase in total root P content (Fig. 6A). Interestingly, while the WT showed higher (P<0.05) 9 shoot P content than root, there was no difference in P content between the two organs in the 10 Pht1;5-overexpression lines. This suggests altered P distribution between source and sink in the 11 overexpressors ( Fig. 6B). It is plausible that Pht1;5-overexpression results in recycling of Pi 12 from the shoots back to the roots. Elevated (~60%) root 33 Pi uptake rates in the Pht1;5-13 overexpressors compared to WT (Fig. 6C) further supported this notion. Although the 33 Pi 14 uptake rates in roots of the Pi-deprived Pht1;5-overexpressors were attenuated compared to WT, 15 the P contents were comparable (data not shown). This indicates a differential effect of Pht1;5-16 overexpression on Pi homeostasis under different Pi regimes. 17 Pht1;5-overexpressors that were grown to maturity under greenhouse conditions 18 consistently displayed chlorosis in older leaves that subsequently senesced earlier than those of 19 WT (Fig. 7A). To gain insight into this premature senescence, 33 Pi tracer studies were carried out 20 to compare the Pi accumulation among a Pht1;5-overexpressor (11C), the pht1;5-1 mutant, and 21 WT (Fig. 7B). Compared to WT,33 Pi accumulation in the Pht1;5-overexpressor was 22 significantly (P<0.05) lower in rosette leaves, comparable in cauline leaves, marginally higher in 23 floral stalks, and ~2-fold higher in siliques. The 33 Pi distribution in pht1;5-1 tissues was 24 consistently and significantly (P<0.05) lower relative to WT. The data suggest increased 25 mobilization of Pi in the Pht1;5-overexpressor from source (i.e. rosette leaves) to sink (i.e. 26 siliques), which may be attributed to enhanced Pi scavenging mechanisms that facilitate release 27 of Pi from bound sources such as nucleic acids (Green, 1994;Raghothama, 1999;Misson et al., 28 2005). qRT-PCR analysis was used to quantify the transcript levels of ribonuclease (RNS1; 29 Bariola et al., 1994) and phosphatase (ACP5; del Pozo et al., 1999) genes involved in Pi 30 remobilization in leaves of four-week-old Pht1;5-overexpressors grown under Pi-replete 31 conditions. As anticipated, there were increases in the relative transcript levels of RNS1 (14 to 28 1 fold) and ACP5 (1.5 to 2 fold) in the Pht1;5-overexpressors compared to WT (Fig. 7C). These 2 molecular data support a role for Pht1;5 in facilitating remobilization of Pi from senescing to 3 metabolically active parts of the plant. 4 5 Pht1;5-overexpression alters root hair development and primary root growth in association 6 with ethylene signaling 7 To study whether differential nutrient partitioning between root and shoot affects root 8 development in the Pht1;5-overexpresion lines, different root traits of the overexpressors were 9 compared with those of WT and the pht1;5-1 mutant. Among the root traits examined, increased 10 root-hair proliferation is an early response of plants to Pi deficiency (Bates and Lynch, 1996;Ma 11 et al., 2001;Jain et al., 2007b). Irrespective of the Pi regime, the Pht1;5-overexpressors showed 12 significant (P<0.05) increases in both the number and length of root hairs compared to WT and 13 pht1;5-1 (Fig. 8). A second Pi-deficiency-induced root response is a reduction in primary root 14 growth due to premature cell differentiation (López-Bucio et al., 2002, 2003Sánchez-Calderón 15 et al., 2005;Jain et al., 2007a). Under P+ and P-conditions, there were significant (P<0.05) 16 reductions in the primary root length of both Pht1;5-overexpressors compared to WT and 17 pht1;5-1 (Fig. 9, A and B). Pi-deficiency-mediated root proliferation aimed at enhancing Pi 18 mining capacity results in an increased root/shoot biomass ratio (Hermans et al., 2006;19 Hammond and White, 2008). Analysis of this trait revealed a significant (P<0.05) reduction in 20 the Pht1;5-overexpressors relative to WT and pht1;5-1 under P-conditions, whereas a reverse 21 trend was observed under P+ conditions ( and ACC + Ag + (Fig. 10). The effects of these treatments on root hair development and primary 4 root length were documented after 1 and 7 d, respectively. Treatment with AVG resulted in 5 significant (P<0.05) reductions in total root hair length of WT and the Pht1;5-overexpressors 6 ( Fig. 10B). However, Ag + treatment resulted in a substantial decrease in total root hair length for 7 the Pht1;5-overexpressors, but only a marginal decrease for WT (Fig. 10B). A similar 8 phenomenon was observed with the impact of AVG on primary root growth. Treatment with 9 AVG had no effect on WT primary root length, but it dramatically increased (P<0.001) primary 10 root growth in the overexpressors, resulting in lengths comparable to WT ( Plants possess multiple Pi transport systems, regulated at both high-and low-affinities, 21 that facilitate Pi uptake from the rhizosphere and subsequently distribute it to cells and 22 subcellular compartments. The movement of Pi inside the plant is an intricate and complex 23 process since source and sink relationships are constantly changing depending on the rate of 24 growth, stage of development, and Pi availability (Raghothama, 1999;Bucher et al., 2001;Miller 25 et al., 2009). The data presented here demonstrate that Pht1;5 is required for normal mobilization 26 of Pi between source and sink tissues. 27 28

Pht1;5 influences shoot Pi status by regulating Pi distribution between root and shoot 29
Among the members of the Pht1 family, Pht1;5 was shown to be expressed in cotyledons 30 and hypocotyls of germinating seedlings, floral buds, older leaves at the onset of senescence, and 31 in Pi-starved roots (Mudge et al., 2002). Our experiments corroborated these results, and also 1 showed that Pht1;5 is expressed in leaves under both P+ and P-conditions (Fig. 1A). Functional 2 inactivation of this gene in pht1;5 mutant seedlings (pht1;5-1 and pht1;5-2) grown under Pi-3 deficient conditions led to low shoot P content, despite also resulting in an increased root Pi 4 uptake rate (Figs 2 and 3). Shoots of mature Pi-starved pht1;5-1 plants showed a similar 5 reduction in P content compared to WT (Supplemental Fig. S1). Decreased shoot P content of 6 pht1;5 mutants correlated with the up-regulation of several Pi-starvation-response (PSR) genes 7 (At4, Pht1;4, DGD1, and SQD1; Fig. 2C). Together these results indicate that the loss of Pht1;5 8 affects mobilization of Pi between root and shoot during Pi deficiency. Pht1;5 may play a role 9 similar to Pht1 transporters from rice (OsPht1;2 and OsPht1;6) and barley (HvPht1;6), which 10 have been implicated in Pi mobilization (Rae et al., 2003;Ai et al., 2009) pht1;5 mutants exhibit an increase (Fig. 3A). This supports the notion that Pht1;4, but not Pht1;5, 28 contributes significantly to root Pi acquisition. Nevertheless, loss of Pht1;5 results in moderate 29 tolerance to arsenate (Fig. 4) For the most part, the two independent T-DNA insertion lines, pht1;5-1 and pht1;5-2, 1 showed similar physiological traits (Figs. 2 and 3). However, distribution of 33 P in the shoots of 2 the mutants differed ( Fig. 2A). This may be due to the presence of multiple T-DNA copies in 3 pht1;5-2 as estimated via qPCR (Supplemental Table S2). Thus, pht1;5-1 was characterized 4 more extensively in this study. 5 Additional evidence supporting the role of Pht1;5 in Pi mobilization came from analysis 6 of plants overexpressing Pht1;5. Under Pi-replete conditions, Pht1;5-overexpressors 7 accumulated more P in roots, but less in shoots, relative to WT (Fig. 6B), which is the opposite 8 trend as that observed for the pht1;5-1 mutant (Fig. 3C). Interestingly though, despite reduced 9 shoot P concentrations the Pht1;5-overexpression lines showed increased shoot biomass (Fig. 5). 10 Our results suggest that overexpression of Pht1;5 results in an imbalance of Pi supply and 11 demand thereby affecting root/shoot biomass allocation ( overexpressors displayed premature leaf senescence (Fig. 7A). Radiotracer experiments of 26 feeding 33 Pi to mature plants revealed that the Pht1;5-overexpressors accumulated 2-fold higher 27 levels of 33 Pi in siliques relative to WT, while the accumulation in rosette leaves was lower (Fig.  28 7B). The Pht1;5-overexpressors also had increased transcript levels for RNS1 and ACP5 in 29 rosette leaves (Fig. 7C). During Pi limitation and senescence, scavenging processes involving the 30 activation of these genes facilitate the release of Pi from organic forms of P (phosphate-31 esters/nucleic acids) thereby making it available for recycling (Bariola et al., 1994;Green, 1994;1 del Pozo et al., 1999). The localization of Pht1;5 in the phloem of older leaves substantiates its 2 role in Pi remobilization (Mudge et al., 2002). Together the data support a role for Pht1;5 in 3 facilitating remobilization of Pi released by scavenging processes from senescing, source (rosette 4 leaves) to metabolically active sink (siliques) organs. A QTL mapping study for seed nutrient 5 content performed using two recombinant inbred populations, Columbia (Col) × Landsberg 6 erecta (Ler) and Cape Verde Islands (Cvi) × Ler, identified Pht1;5 as a putative locus that 7 governs seed P content (Waters and Grusak, 2008). The role of Pht1;5 in loading Pi into seeds 8 warrants further investigation. 9 10 The regulation of Pht1;5 and its interactions with ethylene-and senescence-related 11 pathways 12 Modulation of root system architecture (RSA) to increase root surface area for enhancing 13 Pi adsorption is a characteristic adaptive response of Pi-starved plants (Jain et al., 2007b;14 Schachtman and Shin, 2007;Rouached et al., 2010). Constitutive Pht1;5-overexpression caused 15 increases in the number and length of root hairs, as well as a reduction of primary root length 16 under both P+ and P-conditions (Figs 8 and 9). Several lines of evidence suggest that the 17 modulation of Pi-deficiency-induced changes in RSA result from cross-talk among sugar, 18 nutrient, and phytohormone signaling (López -Bucio et al., 2003;Jain et al., 2007b;Rubio et al., 19 2009). Thus, the alteration of root growth in the Pht1;5-overexpressors could be regulated 20 independently of the Pi regime. Previous studies have implicated ethylene in Pi-deficiency-21 induced root hair formation and primary root elongation (Ma et al., 2001(Ma et al., , 2003Zhang et al., 22 2003;Lei et al., 2011). Ethylene has also been implicated in tolerance to potassium deprivation 23 and iron signaling responses (Zaid et al., 2003;Shin and Schachtman, 2004;Jung et al., 2009). 24 We found that treatment with AVG, an inhibitor of ACC synthase and ethylene biosynthesis, 25 rescued the primary root phenotype of Pht1;5-overexpressors (Fig. 10, A and C), while Ag + 26 treatment decreased the total length of root hairs of Pht1;5-overexpressors to near WT-levels 27 (Fig. 10B). Expression of PhPT1, a high-affinity Pi transporter in petunia, was shown to be 28 shoot by negatively regulating the expression of PHO1, which encodes a protein involved in 1 loading Pi into root xylem (Poirier et al., 1991;Chen et al., 2009). One characteristic feature of 2 WRKY proteins is their ability to autoregulate their own promoter as well as cross-regulating 3 other WRKYs (Rushton et al., 2010). It is possible that feed-forward or feed-back regulation of 4 these determinants could occur concomitantly with the alteration in shoot Pi levels due to the 5 activity of Pht1;5. More experiments are required to dissect the molecular mechanisms that 6 regulate Pht1;5 and its responses that trigger senescence-regulated processes. 7 In conclusion, our data indicate that Pht1;5 plays a complex role in differential partitioning of 8 Pi to plant organs, thereby influencing overall plant growth. Under Pi-replete conditions, Pht1;5 9 expression is limited to shoots and is modulated by developmental cues to promote Pi 10 distribution to sink tissues, including roots of young seedlings and reproductive tissues in mature 11 plants. Upon perception of low Pi, Pht1;5 expression is induced in both shoots and roots. Under 12 these conditions, Pht1;5 is required for proper Pi translocation from root to shoot, likely via a 13 role in loading Pi into root xylem. Pht1;5 also indirectly influences Pi acquisition by regulating 14 shoot P status. agar (Sigma A1296, lot no. 096K01581 with a P content of ~ 30 µM). Seedlings were then 29 transferred to a growth room set to these conditions: 22°C, 16-h photoperiod and average 30 photosynthetically active radiation (PAR) between 60-70 µmol m -2 s -1 . After 5 d, uniformly grown seedlings with primary root lengths ranging between 15-25 mm were transferred to P+ 1 (1.25 mM KH 2 PO 4 ) or P-(P-, 0 mM KH 2 PO 4 ) medium as described (López-Bucio et al., 2002;2 Jain et al., 2009). Although hydroponic conditions are best suited for studying nutrient starvation 3 stress conditions, we observed that lack of added Pi was detrimental to seedling growth and led 4 to eventual death before the completion of the experiment. Therefore the P-treatments were 5 supplemented with 10 μM KH 2 PO 4 for the seedlings to deplete during the course of the 6 experiment. In the Pi-deprived medium, KH 2 PO 4 was replaced by equimolar amount of K 2 SO 4 . 7 Ethylene responsive root growth assays on seedlings were performed using ethylene 8 precursor, 1-aminocyclopropane-1-carboxylic acid (ACC) with or without the addition of 9 aminoethoxyvinyl-glycine (AVG) and silver nitrate (AgNO 3 ) as described in Ruzicka et al., 10 (2007). For overexpression of Pht1;5, the 1.6 kb Pht1;5 cDNA was reverse-transcribed with Super 1 Script III (Invitrogen) from total RNA extracted from Pi-starved Arabidopsis seedlings, and the 2 primers PHT5_S1 (5'-TACGTCGGGCCCCCATGGCGAAAAAAGGAAAAGAAGT-3') and 3

RNA Expression Analyses 12
Total RNA was isolated from ground plant tissues using the RNeasy Plant Mini Kit (Qiagen 13 Inc., www.qiagen.com). One microgram of total RNA was treated with RQ1 RNase-free DNase I 14 (Promega Inc., www.promega.com) and was reverse transcribed using Superscript III TM Reverse 15 Transcriptase (Invitrogen Inc., www.invitrogen.com). Reverse-transcription PCR (RT-PCR) was 16 performed on 2 µL of the cDNA using gene-specific primers. Thermal cycling consisted of an 17 initial denaturation at 94°C for 2 min, followed by 30 cycles (30 s at 94°C, 30 s at 60°C, 90 s at 18 72°C) and final 7 min extension at 72°C. Quantitative RT-PCR analysis was performed on an 19 Applied Biosystems 7500 real-time PCR system using SYBR Green detection chemistry 20 (Applied Biosystems Inc., www.appliedbiosystems.com). UBC and At4g26410 (Czechowski et 21 al., 2005) were used as reference genes, and relative expression levels of the genes were 22 computed by the 2 -∆∆Ct method of relative quantification (Livak and Schmittgen, 2001). In Fig.  23 1A, transcript abundances of Pht1;4 and Pht1;5 across different tissues were calculated as 2 -∆Ct . 24 The primers used are listed in Supplemental Table S1. 25 26

Estimation of T-DNA copy number 27
T-DNA copy number in the genomes of the pht1;5 mutants was estimated via quantitative 28 PCR as described previously by Mason et al., (2002). QuickExtract Plant DNA Extraction 29 Solution (Epicentre, Madison, WI) was used to obtain genomic DNA. Reactions for two 30 biological replicates per genotype were performed in triplicate on an Applied Biosystems 7500 31 real-time PCR system using SYBR Green detection chemistry (Applied Biosystems Inc., 1 www.appliedbiosystems.com). Starting quantities for two regions of the SALK transgene, nptII 2 and a portion near the left border, and two endogenous single-copy genes, Actin 2 and Pht1;4, 3 were determined by LinRegPCR software (Ramakers et al., 2003). For each transgene a virtual 4 calibrator (Masson et al., 2002) was created to use for normalization of the data to represent T-5 DNA copy number. medium for 5d and transferred to P+ (1.25 mM Pi) and P-(0.01 mM Pi) media for 7d. The P+ 20 and P-seedlings along with the meshes were transferred to Petri plates (100 mm x 15 mm) 21 containing P+ or P-nutrient solution supplemented with 0.15 µCi ml -1 [ 33 P] orthophosphate. 22 After 2h, samples were incubated in ice-cold desorption medium (0.1 mM CaCl 2 , 5 mM MES, 23 and 2 mM KH 2 PO 4 , pH 5.7) for 30 min. The samples were rinsed with desorption solution 24 followed by distilled water and blotted dry before separating roots and shoots. After their fresh 25 weights were recorded, tissues were dried overnight at 65 o C, and 33 Pi activity was measured by 26 using a liquid scintillation counter (Tri-Carb 2810, PerkinElmer). The root 33 Pi uptake rate was 27 calculated as pmol Pi g fresh weight -1 hour -1 and distribution was calculated as as pmol Pi g fresh 28 weight -1 . 29 For 33 Pi feeding and tracing experiments, WT, the pht1;5-1 mutant, and Pht1;5-30 overexpressor (11C) were grown on agar-solidified one-half-strength MS medium for 10 d and 31 transferred to well-aerated tubes containing one-fifth-strength Hoagland's nutrient solution 1 containing 100 µM Pi. After 10 d, roots were immersed in the same nutrient solution 2 supplemented with 0.15 µCi ml -1 [ 33 P] for 2 h. Roots were desorbed with unlabelled Hoagland's 3 solution for 30 min and rinsed with tap water thoroughly before transferring back to the nutrient 4 solution. Plants were allowed to grow for 7 d before harvesting different parts of the plant. 5 Tissue samples were dried overnight at 65 o C, weighed, and 33 Pi activity was measured as 6 mentioned above. 7 8 Analysis of RSA 9 Root traits were documented as described (Jain et al., 2007a). Briefly, the seedlings grown on 10 agar-solidified medium were scanned at 600 dpi using a desktop scanner (UMAX PowerLook 11 III) and transferred to 70% (v/v) ethanol facilitating preservation for subsequent detailed 12 analyses. The scanned images were used for measuring the elongation of the primary root after 13 transfer from one-half-strength MS to P+ and P-media. For documenting the number of first-and 14 higher-order lateral roots and their lengths, seedlings stored in ethanol were gently spread on 15 agar (1%, w/v) plates with a camel hair brush and scanned at 600 dpi. 10-15 scanned images of 16 the seedlings per genotype for every treatment were analyzed using ImageJ software 17 (http://rsbweb.nih.gov/ij/). 18

Documentation of Root Hairs 20
Root hair numbers and lengths were recorded as described (Jain et al., 2007a). Briefly, 21 seedlings were initially grown on one-half-strength MS medium with agar (1.2% w/v) for 5d and 22 transferred to agar-solidified P+ and P-media. After 2 d or as mentioned images of root hairs 23 growing in the 5-mm region from the tip of the primary root were captured using a 24 stereomicroscope (Nikon SMZ-U). Root hairs were measured for ~10 seedlings per genotype for 25 every treatment using ImageJ software. 26 27

Statistical Analyses 28
Data were analyzed by one-way (Figs 2A, 3A, 5B-D, 6B-C) or two-way ANOVA (Figs 2B, 3B-29 C, 4B, 6A, 7B, 8B-C, 9B-C, 10B-C) and Tukey's Honestly Significant Difference (HSD) test 30 was carried out for multiple comparisons using the SPSS 10 program (www.spss.com). Different 1 letters represent means that were statistically different at P < 0.05 or as mentioned. inserts (  and mutants were grown hydroponically on one-half-strength MS medium for 5 d and transferred 2 to high (P+) or low (P-) Pi medium for 7 d. A, Rate of Pi uptake in the roots (n = four replicates 3 of ~50 seedlings each). Seedlings grown under P+ and P-were transferred to respective medium 4 supplemented with Pi for 2 h and roots were analyzed for Pi uptake. B, Total P content in the 5 roots (n = four replicates of 25 seedlings each). C, Distribution ratio of P between root and shoot 6 (n = four replicates of 25 seedlings each). Histograms show the means, and different letters 7 indicate means that differ significantly (P<0.05). Error bars represent standard error of the mean. replicates of 20 seedlings each). C, Rate of 33Pi uptake in the roots. Seedlings grown initially 30 under P+ medium were transferred to the same media supplemented with 33Pi for 2 h and roots 31 were analyzed for 33Pi uptake (n = 4 replicates of ~50 seedlings each). Histograms are means, 32 and different letters indicate means that differ significantly (P<0.05). Error bars represent 33 standard error of the mean. supplemented with 33Pi for 2 h. Roots were rinsed with unlabelled nutrient solution and plants 41 were transferred back to the nutrient solution for 7 d before scintillation counting (n = 5). 42 Histogram bars represent means, and different letters indicate means that differ significantly 43 (P<0.05). Error bars indicate standard error of the mean. C, Quantitative RT-PCR analyses 44 showing increased expression of genes involved in Pi remobilization in Pht1;5-overexpressors. 45 Three-week-old plants were grown hydroponically in modified one-half-strength Hoagland's 46 nutrient solution for 7 d and transferred to high Pi media for 7 d. At4g26410 was used as an 1 internal control and the values, normalized to WT levels, are the RQ ± maximum/minimum 2 values of two independent biological replicates run in triplicate. Asterisks (*) and (**) indicate 3 ≥ 1.5-fold and ≥ 2.0-fold changes, respectively, compared to WT. 4 5 Figure 8. Pht1;5-overexpression leads to increases in root hair number and length. WT, pht1;5-6 1, and Pht1;5-overexpressors (5A and 11C) were grown on agar-solidified one-half-strength MS 7 medium for 5 d and transferred to high (P+) or low (P-) Pi medium for 2 d. A, Primary root tips 8 of the representative seedlings showing root hairs (n = 10). Data are presented for numbers (B) 9 and total lengths of the root hairs in a 5-mm section from the primary root tip (C). Histograms 10 are means, and different letters indicate means that differ significantly (P<0.05). Error bars 11 represent standard error of the mean. 12 13 Figure 9. Pht1;5 loss-of-function and overexpression affects root system architecture. WT, 14 pht1;5-1, and Pht1;5-overexpressors (5A and 11C) were grown on agar-solidified one-15 halfstrength MS medium for 5 d and transferred to high (P+) or low (P-) Pi medium for 7 d. A, 16 Root system architecture of the representative seedlings (n = 12). B, Primary root length (n = 17 12). C, Root-shoot biomass ratio recorded on a dry weight basis (n = 5 replicates of 20 seedlings 18 each). B and C, Histograms are means, and different letters indicate means that differ 19 significantly(P<0.05). Error bars represent standard error of the mean. 20 21 Figure 10. Ethylene signaling and biosynthesis inhibitors rescue the root phenotype of Pht1;5-22 overexpressors. WT and the Pht1;5-overexpressors (5A and 11C) were grown on agar-solidified 23 one-half-strength MS medium for 5 d and then transferred to P+ medium (control), or the same 24 medium supplemented with the inhibitors shown, for 7 d. A, Phenotype of WT and 11C 25 seedlings grown in the presence of indicated treatments. B, Total root hair length in a 5-mm 26 section from the primary root tip (n = 10). C, Primary root length under the indicated treatments 27 (n = 16). Histograms are means, and different letters indicate means that differ significantly 28 (P<0.05). Asterisks (**) represent means that differ significantly at (P<0.001) compared with the 29 untreated control sets. 30 Figure 1. Spatial expression pattern of Pht1;5 and isolation of pht1;5 loss-of-function mutants. A, Quantitative RT-PCR analysis of the expression of Pht1;4 and Pht1;5 in different organs under different Pi regimes. WT and the mutants were raised hydroponically under aseptic conditions on one-half-strength MS medium for 5 d and transferred to high (P+) or low (P-) Pi medium for 7 d. B, Schematic representation of the Pht1;5 gene and the location of T-DNA inserts (filled triangles) in the different pht1;5 alleles. Open and filled boxes represent untranslated regions and exons, respectively. C, Reverse-transcription PCR analysis of Pht1;5 expression in the WT and pht1;5 mutants. Seedlings were grown hydroponically under aseptic conditions on one-half-strength MS medium for 5 d and transferred to P-medium for 7 d. Whole seedlings were harvested. Transcript abundance was determined using primers specific to Pht1;5 and UBC genes.