Plants display a range of adaptive responses to phosphate (Pi) starvation including an increase in the proportion of Pi allocated to the roots, which enhances lateral root development and consequently Pi acquisition. The mechanisms by which plants sense Pi and signal Pi reallocation are largely unknown. Previously, we cloned At4, a gene predicted to contain multiple short open-reading frames (ORFs), whose expression is strongly induced by Pi starvation. At4 is a member of a small gene family whose members, AtIPS1 and two additional genes reported here, At4.1 and At4.2, share little conservation among the predicted ORFs but high conservation of a 22-nt sequence located in the 3′ half of the transcript. Here, we show that under Pi-starvation conditions, At4 is expressed in the vascular tissue and transcript levels are regulated by both cytokinin and ABA. at4, an At4 loss-of-function mutant fails to redistribute Pi to the roots correctly in response to Pi deprivation and At4 shoots continue to accumulate a greater proportion of Pi relative to wild type. Consistent with this, the primary root growth rate in at4 is faster than wild type in low-Pi conditions. The conserved sequence found in all members of the At4 gene family hybridizes to a small RNA present in Pi-starved roots. These data support a role for At4 in the internal allocation of Pi and suggest that the At4 gene is not only subject to Pi-starvation-inducible expression, but that transcript levels may be adjusted at a post-transcriptional level by the activity of an miRNA.
Plants require a number of mineral nutrients for growth and development (Marschner, 1995). Of the essential macronutrients, phosphorus (P) is one of the most important as it plays indispensable roles in major metabolic processes, including photosynthesis, energy transfer reactions and carbon metabolism. Furthermore, it is a structural component of cellular macromolecules and a key regulator of enzymes and signalling cascades. Phosphorus comprises approximately 0.2% of the dry weight of the plant (Schachtman et al., 1998). Consequently, plants must not only obtain considerable amounts of P from the soil but must coordinate its allocation throughout the plant to meet the growth and metabolic requirements of each tissue. Over the last few years, there has been significant progress in the identification of the Pi transport systems that operate to acquire Pi from the soil and mediate its movement to the vascular system, aerial tissues and various organelles, including the chloroplast (Chiou et al., 2001; Daram et al., 1998, 1999; Hamburger et al., 2002; Liu et al., 1998a,b; Muchhal et al., 1996; Mudge et al., 2002; Paszkowski et al., 2002; Shin et al., 2004; Versaw and Harrison, 2002; Wang et al., 2004).
Soils are heterogeneous environments in which P availability varies considerably. Although P is abundant in soils, only a fraction exists in the ionic forms H2PO and HPO, that are available to plants, and it is one of the mineral nutrients that most frequently limits plant growth (Bieleski, 1973; Holford, 1997; Vance, 2001). As might be expected for such a critical element, plants have evolved strategies both to exploit localized sources of phosphate (Pi) and to cope with Pi deprivation (Drew, 1975; Robinson, 1994; Vance et al., 2003). During periods of Pi deprivation, plants undergo biochemical changes that reduce the Pi requirement including the use of alternate reactions in the respiratory and glycolytic pathway, alterations in lipid composition and carbohydrate metabolism (Kelly and Dormann, 2004; Theodorou and Plaxton, 1993; Ticconi and Abel, 2004). They also implement a range of strategies designed to enhance Pi acquisition. For example, there are significant shifts in the allocation of Pi within the plant with the net effect of increasing Pi allocation to the roots and concomitant increases in root biomass relative to the shoot biomass (Jeschke et al., 1996, 1997; Mimura, 1999). Enhanced root growth is accompanied by alterations in root architecture, including a reduction in primary root growth and increases in lateral root growth, and in the number and length of root hairs (Bates and Lynch, 1996; Drew, 1975; Lopez-Bucio et al., 2002; Williamson et al., 2001). In addition, expression of enzymes and proteins involved in Pi acquisition, such as secreted phosphatases (Baldwin et al., 2001; Bozzo et al., 2002; Goldstein et al., 1988), RNAses (Abel et al., 2000; Bariola et al., 1994) and Pi transporters increases (Raghothama, 1999; Rausch and Bucher, 2002). Another Pi acquisition strategy, widely employed by vascular flowering plants, is the formation of symbiotic associations with arbuscular mycorrhizal (AM) fungi, which assist the plant with the acquisition of Pi from the soil (Harrison, 1997). In those plant species that do not form AM symbioses, other adaptations, such as proteoid root formation, enable the plant to adapt to low-Pi environments (Liu et al., 2005).
Recent genome-scale transcriptional profiling has revealed a myriad of genes whose expression is altered during Pi deprivation (Hammond et al., 2003; Wu et al., 2003). The function of many of these genes remains to be elucidated but these analyses provide useful targets, which can serve as starting points for the dissection of the Pi-starvation response. The mechanisms via which these genes are regulated is largely unknown; however, expression of at least one subset of Pi-starvation-inducible (PSI) genes is regulated by a transcription factor, PHR1 (Pi starvation response; Rubio et al., 2001).
Relatively little is known about how plants sense Pi but various physiological responses, such as local alterations in root hair number and length during growth on low-Pi media, indicate that plants roots can sense and respond to Pi in the local external environment (Abel et al., 2002). An Arabidopsis mutant, pdr1, is defective in a Pi-sensing mechanism that controls lateral root meristem number and may encode a component of a local Pi-sensing pathway (Ticconi et al., 2004). However, local Pi sensing is only one part of the system and some responses, including primary root growth and expression of Pi-starvation-inducible genes, are controlled systemically in response to the Pi status of the shoot, indicating that the Pi status of the whole plant is sensed also (Abel et al., 2002). The Arabidopsis pho2 mutant, accumulates up to five times more Pi in the shoots relative to wild type (Delhaize and Randall, 1995; Dong et al., 1998) and may encode a component of a shoot Pi-sensing system or the system that controls the allocation of Pi between shoots and roots.
Previously, we identified Mt4 from Medicago truncatula and its Arabidopsis, ortholog, At4. Both genes are rapidly induced following Pi starvation (Burleigh and Harrison, 1998a,b, 1999). In other studies Pi-starvation-inducible At4/Mt4-like genes from tomato (TPSI1; Liu et al., 1997), Arabidopsis thaliana (AtIPS1; Martín et al., 2000), and rice (OsPI1; Wasaki et al., 2003) were identified and this family of transcripts has been referred to as the Mt4/TPSI1family. The functions of these genes are unknown but their transcripts accumulate to high levels in Pi-starved tissues and, consequently, they have been used widely as molecular indicators of Pi starvation (Burleigh and Harrison, 1998b; Shin et al., 2004; Ticconi et al., 2004). Transcripts from the Mt4/TPSI1 genes lack a single long open-reading frame (ORF). Instead, they contain a series of short overlapping ORFs that are not conserved among the family members. Overall, the sequence identities among these genes are relatively low except for 22 nt in the central region of each transcript that is highly conserved.
Here, we demonstrate a role for At4 in the allocation of Pi between shoot and roots and present evidence of a small Pi-starvation regulated RNA complementary to the At4 sequence.
Identification of At4.1 and At4.2, two additional members of the At4 gene family
At4 was identified as a potential ortholog of a Pi-starvation-inducible gene, Mt4 of M. truncatula (Burleigh and Harrison, 1999). The subsequent identification of a related sequence, AtIPSI (Martín et al., 2000) prompted us to search for additional members of this gene family. Using a gene-specific primer designed to the conserved region of the Mt4/At4 genes and RT-PCR, we identified two new At4-like transcripts from A. thaliana. The two genes were designated At4.1 and At4.2. Sequence comparisons indicate that At4.1 and At4.2 show little sequence identity with At4 or Mt4, except for a region of 22 nt in the middle of each transcript that is highly conserved in all Mt4/At4 family members (Figure 1a). Like the At4 gene, At4.1 and At4.2 transcripts contain numerous, short, overlapping ORFs that could encode small peptides. The peptides predicted from these ORFs are not conserved among Mt4/At4 gene family members. The longest ORFs from At4.1 and At4.2 are predicted to encode peptides of 62 and 25 amino acids, respectively.
blast analyses identified at least three expressed sequence tags (ESTs) (GeneBank ID: AU236364, AU227270, AA597438) corresponding to the At4.1 gene but no ESTs corresponding to At4.2. Analysis of the Arabidopsis genome sequence revealed that At4.1 and At4.2 were found in regions designated as intergenic, indicating that the current gene prediction algorithms did not recognize these as genes. At4.1 was located in the intergenic region between At1g07770 and At1g07760, and At4.2 between At2g36730 and At2g36720. At4.1 and At4.2 do not contain introns, another common characteristic of the Mt4/At4 gene family.
Expression of members of the Arabidopsis At4 gene family is regulated by Pi availability
The expression of At4, At4.1 and At4.2 was compared by Northern blot analysis with total RNA from plants grown in sand under low- and high-Pi conditions (Figure 1b). As expected, At4 expression was induced strongly in roots of plants grown under low-Pi conditions. As shown previously, Pi availability regulates the expression of At4 and starvation for other nutrients such as sulphate and nitrate exerted no detectable effect on At4 expression. At4.1 and At4.2 showed similar expression patterns to At4; however, the transcript levels of At4.1 and At4.2 were much lower than that of At4 (Figure 1b,c; note the figure legend for details of the exposure times).
To examine the effects of Pi availability on expression of At4, At4.1 and At4.2 in more detail, we carried out Northern blot analysis using total RNA isolated from the roots of Arabidopsis grown in sand and fertilized with increasing levels of Pi (Figure 1d). At4 transcript levels were highest in plants fertilized with 0.1 mm Pi and, although transcript levels decreased with increasing Pi fertilization, they were still readily detected in the roots of plants fertilized with up to 0.4 mm Pi. In contrast, A4.1 and At4.2 transcripts were detected in plants fertilized with 0.1 mm Pi and almost undetectable from plants fertilized at 0.2 mm Pi or higher. Based on the Northern analyses, the transcripts of the At4.1 and At4.2 genes are slightly smaller than that of At4 and are estimated to be approximately 600 nt.
Cytokinin and absiscic acid reduce At4, At4.1 and At4.2 transcript levels during Pi starvation
It has been shown previously that cytokinin negatively regulates expression of At4 and AtIPS (Martín et al., 2000). We investigated the effect of 1-aminocyclopropane-1-carboxylic acid (ACC), salicylic acid (SA), indole-acetic acid (IAA), abscisic acid (ABA), kinetin, gibberellin (GA3), and aminoethoxyvinylglycine (AVG)on the expression of At4 and showed that, in addition to kinetin, ABA treatment could also repress the expression of At4 in the Pi-starved roots. The other hormones had no detectable effect on At4 transcript levels (data not shown). We extended these analyses to include At4.1 and At4.2. As shown in Figure 1(e), compared with the untreated controls, At4, At4.1 and At4.2 transcripts were significantly lower in Pi-starved plants treated with ABA or kinetin, while AVG application did not have any visible effect on the transcript levels of these genes.
The At4 promoter drives expression of a reporter gene in a Pi-starvation-dependent manner
Among the Arabidopsis At4 gene family members, At4 shows the highest transcript levels in Pi-starved roots. To examine the spatial expression of At4, we generated transgenic Arabidopsis plants transformed with a GUS gene fused with 2 kb of genomic DNA from the 5′ proximal region of the At4 gene. Twelve independent T2 lines were obtained, all of which showed GUS activity in response to Pi starvation. Three lines were selected for further analysis.
The transgenic lines (T3 generation) were grown for 2 weeks on agar media containing 0 (low Pi) or 1 mm Pi (high Pi). As shown in Figure 2, when grown in low-Pi conditions, the At4 promoter-GUS transgenic Arabidopsis plants displayed strong GUS activity in roots; however, no GUS activity was detected in the roots of the seedlings grown in high Pi, indicating that the promoter sequence directs expression in a Pi-starvation-inducible manner (Figure 2a,b).
To further analyze the expression patterns, the transgenic lines were grown for 2 weeks on agar media containing 1 mm Pi and then transferred to media containing no external Pi and grown for an additional 1 to 2 weeks. Plants grown for 1 week on growth media lacking Pi showed strong GUS staining in the vascular tissues of the root. Plants grown for 2 weeks on media lacking Pi showed strong GUS staining in the vascular tissue and also in the cortical and epidermal cells in some regions of the root (Figure 2c,d). Under this growth condition, GUS staining was visible in the vascular tissues of the leaves also (data not shown).
Isolation of a mutant with a T-DNA insertion in the At4 gene
Utilizing a PCR-based screen, we isolated an Arabidopsis line, at4 which had a T-DNA insertion in the transcribed region of the At4 gene (Figure 3a,b). Sequence analysis revealed that the T-DNA was located between 539 and 578 nt downstream of the first putative ATG codon of At4 and insertion had resulted in a deletion of 38 nt. A homozygous line was selected and expression of the At4 gene was examined by RT-PCR and Northern blot analysis. The At4 transcript could not be detected in the at4 line by Northern or RT-PCR analysis, regardless of the Pi growth conditions. This indicates that at4 is a null allele (Figure 3c). The expression of At4.1 and At4.2 was not altered in the at4 mutant (Figure 3c).
Pi content and growth of at4
The at4 mutant did not show any obvious visible phenotypic differences relative to wild type when grown in soil. To examine the effects of growth under conditions of Pi deprivation, plants were grown on agar plates containing 500 μm Pi for 2 weeks and then transferred to fresh agar plates containing no additional Pi and grown for an additional 2 weeks prior to harvest. Previous experiments with the At4 promoter-GUS transgenic plants had shown that, under this growth regime, plants show symptoms of Pi starvation, including strong induction of the At4 gene, within 1 week of the transfer to Pi-free media. Under these conditions, at4 and wild-type plants showed similar shoot and root mass; however, at4 showed a small, but significant increase in shoot Pi content relative to the levels in the wild-type plants. The mean shoot Pi content of at4 plants was 17.3% higher than that of the wild-type plants (Table 1). The Pi content of at4 and wild-type roots were not significantly different, but the ratio of Pi in the shoots relative to the roots differed significantly between the at4 and wild-type lines, with at4 showing a mean Pi shoot:root ratio that was 36.1% higher than that of wild type (Table 1). Consistent with a reduction in the Pi content of the roots, at4 shows a small increase in the expression of the major Pi-starvation-inducible Pi transporters of the root system (Figure 4). Pht1;1, Pht1;2, Pht1;3 and Pht1;4 transcript levels are slightly higher in at4 relative to wild type during growth under Pi-starvation conditions. While Pht1;5, Pht1;7, Pht1;8 and Pht1;9 transcripts remain at levels comparable to those seen in wild type (Figure 4).
Table 1. Effect of Pi-starvation on the growth and Pi content of at4. Plants were grown for 10 days on 0.5 × Hoagland agar plates containing 500 μm Pi and were then transferred to 0.5 × Hoagland agar containing no Pi for an additional 2 weeks. Measurements were made on individual plants. The values represent mean ± SE (n = 20 for fresh weight and n = 8 for Pi content). Different letters indicate statistically significant differences (Student's t-test, P < 0.05)
Plant growth (mg FW)
5.80 ± 0.30a
8.86 ± 0.75b
0.71 ± 0.04c
5.73 ± 0.30a
8.33 ± 0.56b
0.72 ± 0.03c
Pi content (nmol Pi mg−1 FW)
Shoot Pi content
Root Pi content
2.25 ± 0.09a
1.25 ± 0.08c
1.83 ± 0.09d
2.64 ± 0.17b
1.10 ± 0.09c
2.49 ± 0.29e
At4 may play a role in regulating Pi distribution between roots and shoots during growth in low-Pi conditions but, under such conditions, when growth rate is slower, the effect of the loss of At4 may be too small to result in significant alterations in growth. To increase the chance of observing changes resulting from the loss of At4 function, we took advantage of a double Pi transporter mutant, pht1;1Δ4Δ, defective in two major root Pi transporters, Pht1;1 and Pht1;4, that we had created previously. The Pi transporter double mutant shows Pi starvation even at high-Pi conditions and At4 is expressed in the double mutant during growth in high-Pi conditions (Shin et al., 2004). Our rationale was that the effect of the loss of At4 might be magnified in this background. at4 was crossed to pht1;1Δ4Δ and a triple mutant at4, pht1;1Δ4Δ was identified from the F2.
In an initial experiment, wild-type and mutant lines were grown for 3 weeks on 0.5 × MS media and the biomass of the plants determined. As expected, under these conditions, at4 shoot mass was not significantly different to wild type, while the double mutant (pht1;1Δ4Δ) showed a significant reduction in growth relative to wild type. In contrast, the average shoot mass of the triple mutant (at4 pht1;1Δ4Δ) was significantly higher than that of the double mutant and showed a 30% increase relative to that of the double mutant (Table S1). Similar results were obtained in a subsequent experiment in which the plants were grown in 0.5 × Hoagland agar medium containing 500 μm Pi and the mass and Pi content of individual plants was assessed (Table 2). Under these high-Pi conditions, the shoot mass and shoot Pi content of at4 and wild-type plants did not differ significantly. However, the mean shoot Pi content of the triple mutant was 49.3% higher than that of the double mutant and the Pi shoot:root ratio was 48.4% higher in the triple mutant compared with the double mutant. This difference in Pi allocation translated to a significant difference in growth also, and the triple mutant showed mean shoot and root fresh weights that exceeded those of the double mutant by 40.1 and 44.9%, respectively. Collectively, these data indicate that At4 function affects the distribution of Pi between the roots and shoots, resulting in an increase in the amount of Pi allocated to the shoots and a concomitant increase in growth.
Table 2. Plant growth in an at4, pht1;1Δ4Δ triple mutant. Plants were grown for 3 weeks on cellophane membranes on Hoagland agar plates containing 500 μm Pi. Fresh weight and Pi content measurements were made on individual plants. The values represent mean ± SE (n = 19 for fresh weight, n = 8 for Pi content). Different letters indicate statistically significant differences (Student's t-test, P < 0.05). The at4, pht1;1Δ4Δ line is referred to as triple
Growth (mg FW)
15.91 ± 1.51a
6.03 ± 0.67c
2.77 ± 0.12e
15.21 ± 1.25a
6.51 ± 0.60c
2.42 ± 0.14e
9.86 ± 0.71b
4.56 ± 0.39d
2.27 ± 0.11f
13.86 ± 1.33a
6.61 ± 0.74c
2.20 ± 0.20f
Pi content (nmol mg−1 FW)
10.52 ± 0.33a
6.80 ± 0.32d
1.57 ± 0.11g
11.61 ± 0.65a
5.54 ± 0.12e
2.11 ± 0.14h
3.41 ± 0.08b
3.80 ± 0.34f
0.93 ± 0.08i
5.09 ± 0.36c
3.62 ± 0.1f
1.38 ± 0.09g
at4 shows a reduction in Pi uptake but an increase in Pi allocation to the shoots
To determine whether the loss of At4 affects Pi transport activities, we carried out Pi uptake experiments. Pools of plants were grown floating on 0.5 × Hoagland media (0.3% agar) containing either 2 or 500 μm Pi. Following 2 weeks of growth, the plants were transferred to Pi uptake solutions containing either 2 or 500 μm Pi supplemented with P33 as a radiotracer and Pi uptake into the roots was measured. There was no significant difference in Pi uptake in the wild-type and at4 plants grown at 500 μm Pi. In contrast, at4 plants grown at 2 μm Pi showed a 37.2% reduction in Pi uptake rate relative to the wild type (100% = 80.9 ± 5.1 nmol Pi g−1 FW h).
To examine the Pi transport and distribution to the shoots, additional experiments were undertaken using individual plants rather than pools. Plants were grown for 2 and 4 weeks on the surface of cellophane membranes supported on agar plates and equilibrated with 0.5 × Hoagland 5 μm Pi media. This enabled individual plants to be removed easily without disturbance or loss of root tissue. For Pi transport studies, plants were removed from the cellophane and incubated in Pi uptake solution containing 5 μm Pi with 1 μCi P33 and the P33 content of the shoots and roots was determined (Figure 5a and Table 3). In this experiment, the difference in the rate of Pi transport into the roots of at4 plants relative to wild-type plants was not statistically significant. However, compared with the wild type, Pi transfer to the shoots was significantly lower in at4 plants harvested at 2 weeks and significantly higher in plants harvested at 4 weeks. In both at4 and wild-type plants, Pi uptake into the roots increased dramatically after 4 weeks of growth at 5 μm Pi (Table 3 and Figure 5a), which is consistent with increased Pi starvation and the activation of Pi-starvation-inducible Pi transporters (Muchhal and Raghothama, 1999; Muchhal et al., 1996; Shin et al., 2004). In the wild-type lines, the Pi shoot:root ratio in the plants harvested at 4 weeks was 5.6-fold lower than those harvested at 2 weeks. A reduction in the allocation of Pi to the shoots and an increase in Pi allocation to the roots is a common physiological response to Pi starvation and this result further indicates that the plants harvested at 4 weeks were experiencing severe Pi deprivation. Interestingly, while the Pi shoot:root ratio in at4 plants also decreased in the 4-week samples relative to the 2-week samples, the decrease was only 2.4-fold and the Pi shoot:root ratio remained significantly higher than that of wild type. This suggests that although the at4 plants harvested at 4 weeks experience severe Pi deprivation, they do not respond in the same way as wild type and continue to translocate a greater proportion of the Pi to the shoots. Decapitation of the seedlings did not affect the Pi uptake into the roots significantly in wild type or at4 (data not shown), indicating that the altered Pi translocation to the shoots observed in the at4 plants is likely due to the modification of the internal translocation activities. These results further indicate that regulation of the Pi translocation to the shoots in at4 is altered and that this is influenced by the Pi-starvation status of the plant.
Table 3. Pi uptake and transfer to shoots in at4. The plants were grown on 0.5 × Hoagland agar containing 5 μm Pi for 2 or 4 weeks. Pi uptake and transfer to the shoots was measured using 33Pi and measurements were made on individual plants (pmol Pi−1 plant−1 h). The values represent the mean ± SE (n ≥ 8). The letters represents statistically significant differences (Student's t-test, P < 0.05)
Pi-absorption rates (2 weeks)
Pi-absorption rates (4 weeks)
41.9 ± 5.5a
113.5 ± 23.4c
0.37 ± 0.03d
23.9 ± 3.4a
383.6 ± 27.3c
0.066 ± 0.012d
23.0 ± 4.1b
93.4 ± 10.4c
0.25 ± 0.04e
35.1 ± 4.0b
334.5 ± 23.4c
0.104 ± 0.019e
The growth rate of the primary root is enhanced in at4 at low Pi
Pi is a key regulator of root system architecture and, generally, low Pi availability results in a reduction in the rate of growth of the primary root and an increase in the lateral root number and growth rate. This process is controlled at least in part, by the Pi content of the shoots (Lopez-Bucio et al., 2002; Williamson et al., 2001). Consistent with this, we observed that the primary roots of wild-type plants showed a reduction in growth rate in low-Pi conditions (Figure 5b). During growth in high-Pi conditions, the primary roots of at4 plants showed similar growth rates to wild type. However, under low-Pi conditions, the primary roots of at4 plants grew 31.7% faster than wild type. This is consistent with the previous data that indicated increased allocation of Pi to the shoots in at4.
Complementation of at4 and functional complementation with the Mt4 gene
The At4 gene and approximately 2 kb of upstream sequence were introduced into at4. Two lines were selected and Pi transport and translocation and primary root growth were assessed. As shown in Table 4, introduction of the wild-type At4 gene complemented the Pi translocation defect observed under low-Pi conditions and the shoot:root Pi absorption rate was restored to wild-type levels. Introduction of the wild-type At4 gene also complemented the primary root growth phenotype and restored primary root growth to wild-type levels (Table S2).
Table 4. Complementation of at4. Plants were grown for 3 weeks on agar containing either no additional Pi (low Pi) or 500 μm (high Pi) Pi. Pi uptake and transfer to the shoots was measured using 33Pi and measurements were made on individual plants (pmol Pi−1 plant−1 h). Ratio of Pi in the shoot to the root (S:R) is displayed. at4:At4#5 and #14 are individual at4 lines transformed with the wild-type At4 gene. at4::35S-Mt4 is the at4 mutant expressing a 35S-Mt4 transgene. Values are the mean ± SE (n = 10). Different letters indicate statistically significant differences (Student's t-test P < 0.05). Ws, Wassilewskija.
S:R ratio of Pi
0.034 ± 0.007a
0.33 ± 0.02a
0.062 ± 0.012b
0.28 ± 0.03a
0.039 ± 0.007a
0.31 ± 0.04a
0.038 ± 0.008a
0.23 ± 0.03a
0.013 ± 0.004c
0.61 ± 0.12b
Although the members of the At4/Mt4 gene family show similar expression patterns and share the unique 22-nt conserved sequence, it is not known whether they also share similar biological functions. As an initial step to address this question, we introduced a full-length Mt4 cDNA sequence under control of the 35S promoter into at4. Under low-Pi conditions, the Mt4 sequence was able to complement the shoot Pi translocation defect and as shown in Table 4, the shoot:root Pi absorption rate was reduced to a level significantly lower than the wild-type level. This might be an effect of constitutive expression of the Mt4 gene. Surprisingly, under high-Pi conditions, constitutive expression of the Mt4 gene in at4 resulted in increased allocation of Pi to the shoots and a significant increase in the ratio of Pi in the shoot versus the roots relative to both the at4 levels and the wild-type controls.
A small RNA expressed during Pi starvation is detected by a 21-nt probe corresponding to the conserved region of At4
Various features of the At4 family genes, including the lack of conservation among the small ORFs and the high conservation at the nucleotide level suggested that they might be non-coding RNAs, whose function is mediated at the RNA level. Among the non-coding RNAs, microRNAs (miRNAs), derived from larger RNA precursors, are emerging as significant regulators of many processes in plants. These ∼22-nt RNAs pair with their mRNA targets and initiate cleavage or inhibit translation (Bartel and Bartel, 2003; Carrington and Ambros, 2003). Computational analysis of the At4, At4.1 and At4.2 sequences suggests that they do not have the potential to form a hairpin structure typical of miRNA precursor molecules (Matthews et al., 1999). Therefore, to explore the possibility that the conserved 22-nt site is the target for an miRNA, we performed Northern blot analyses using size-fractionated small RNAs isolated from root tissues of plants grown under low- and high-Pi conditions. The blots were hybridized with oligonucleotide probes corresponding to the first 21 nt of the At4 conserved sequence in either the sense or antisense orientation. Hybridization with the At4 21-nt sense probe revealed the presence of a small RNA of ∼22 nt in the roots of Pi-starved plants (Figure 5c). This species was not detectable in plants grown with high levels of Pi fertilization. In contrast, a probe antisense to the At4 21-nt sequence did not detect a small RNA species.
This analysis was extended to the at4 mutant and the at4 complemented lines (Figure 6). Plants were grown in 2 mm Pi for the 2 weeks and then transferred to fresh media containing either 2 or 0 mm Pi and grown for an additional week before harvest. As observed previously, the At4 21-nt sense probe detected a small 22-nt RNA in RNA samples isolated from Pi-starved plants. This small RNA was present in at4, which further indicates that it is not a product of the At4 sequence, and was present in the complemented lines also (Figure 6a). Two larger RNA species, of between 75 and 100 nt, were detected also by the sense probe. These larger species were present in both low- and high-Pi conditions and it is possible that they are progenitors of the small RNA. Analyses of the Arabidopsis Small RNA Project (ASRP) database (Gustafson et al., 2005) and a recent report of novel small RNAs from Arabidopsis (Sunkar and Zhu, 2004) led to the identification of a microRNA, miR-399b, that is partially complementary to the At4 22-nt conserved sequence. The miR399 family is a large miRNA family and the small RNA detected on our blots might be a member of this family.
As observed in the initial experiment, the At4 21-nt antisense probe did not detect a small RNA species but hybridized to a diffuse smear of RNA species of less than 120 nt in size from wild-type plants grown in low Pi but these were not detected in plants grown with high Pi. These species were not detected in at4 regardless of Pi conditions, but were present in the at4 line complemented with an At4 genomic fragment during growth at low Pi and not at high Pi (Figure 6b). This result supports the hypothesis that these species are degradation products of the At4 transcript. In further support of this hypothesis, these RNA species were present at both high and low Pi in the at4 line complemented with an Mt4 cDNA under control of the 35S promoter, although the signal at high Pi was considerably weaker, as is the level of the small RNA species. Collectively, these results indicate the presence of a Pi-starvation-inducible small RNA, complementary in sequence to the conserved region found in the At4/Mt4 genes. The data also provide evidence that the At4 and Mt4 transcripts are degraded to the smaller products of less than 120 nt.
In summary, our analyses suggest that At4 plays a role in the Pi allocation within the plant and that At4 transcript levels may be regulated not only by Pi-starvation- induced activation of the promoter, but also post-transcriptionally, by a small RNA.
Pi acquisition by plant roots and its allocation throughout the plant is a finely tuned but dynamic process that enables the plant to meet its needs for growth while maintaining the flexibility to respond rapidly to a changing environment. Physiological studies of Pi transport and allocation suggest that these processes are regulated in response to both local and systemic signals. While many of the transport proteins involved in Pi acquisition and distribution are known, very few of the signalling or regulatory components have been identified (Abel et al., 2002; Ticconi and Abel, 2004).
Members of the Mt4/TPSI1 family were among the first Pi-starvation-inducible genes to be identified and this was undoubtedly because of the extreme abundance of their transcripts in Pi-deprived roots. However, the functions of these genes have remained an enigma with the presence of numerous overlapping ORFs making it difficult to predict the nature of the protein product (Burleigh and Harrison, 1998b; Liu et al., 1997).
At4 was the first member of this family to be identified in Arabidopsis. Subsequently a second member, AtIPS1, was identified and as described here, we have identified two further members of the Arabidopsis family At4.1 and At4.2. The transcripts share relatively little similarity at the nucleotide level except for a conserved 22-nt region in the central portion of the transcript. The transcript levels of all members of the family are elevated significantly in response to Pi starvation and as described for AtIPS1 and, shown here for At4, the promoters of these genes are responsive to Pi deprivation. Analysis of an AtIPS1 promoter GUS fusion construct indicated that AtPSI1 is expressed initially in the elongation zone of the root and in the cotyledons and following further Pi deprivation, throughout the plant, in all cells that show Pi deprivation (Martín et al., 2000). In contrast, At4 is expressed most highly in the vascular tissue of the root and, following severe deprivation, expression extends to other cell types within the root and also to the vascular tissues of the shoot.
Here, we have shown that the loss of At4 function impacts the allocation of Pi between the shoots and the roots during growth in low-Pi conditions. In general, during Pi starvation, plants allocate a greater proportion of Pi to the roots and root growth is enhanced at the expense of shoot growth, a response that enables the plant to maximize further Pi acquisition (Ticconi and Abel, 2004). In at4, this allocation process is altered and, during Pi starvation, the shoots contain significantly more Pi than the comparable wild-type controls. Consistent with the expression patterns of At4, this effect is observed only during periods of Pi starvation. The altered allocation of Pi results in a modest alteration in shoot growth in the at4 mutant relative to wild type but this is magnified significantly in the background of a Pi transporter double mutant. The Pi transporter double mutant lacks Pht1;1 and Pht1;4, the two Pi transporters responsible for the majority of Pi uptake into roots. Consequently, it shows symptoms of Pi deprivation, including expression of At4, even during growth at higher external Pi levels, and is more sensitive to small increases in Pi (Shin et al., 2004). Analysis of a triple mutant lacking At4 and the two Pi transporters, further demonstrated that the loss of At4 function alters the allocation of Pi to the shoots and, in this line, the resulting effects on shoot growth were observed. Relative to the Pi transporter double mutant, the triple mutant allocated 48% more Pi to the shoots and the shoots were 40% larger than the corresponding double mutant control line. Experiments in this line also indicate that the effect is not influenced by the Pi concentration of the external environment, but is likely under the control of internal Pi levels.
The Pi allocation phenotype of the at4 mutant was complemented by introduction of a genomic copy of the At4 gene and, interestingly, constitutive expression of Mt4, the putative ortholog from M. truncatula, complemented the mutation also and actually shifted the balance of Pi allocation further in favour of the roots. It is possible that this is a result of constitutive versus regulated expression. This result suggests that Mt4 is a functional ortholog of At4. Surprisingly, constitutive expression of Mt4 during growth in high-Pi conditions resulted in the opposite effect and the Pi shoot:root ratio was two times higher than wild type. Again, this might be an effect of constitutive expression as At4 would not normally be expressed during growth in high-Pi conditions.
Pi is translocated from the roots to the shoots in the xylem and PHO1, a protein located in the vascular tissue, is involved in this process (Hamburger et al., 2002; Poirier et al., 1991). During Pi starvation, Pi translocation to the shoots continues but a greater proportion of Pi is retranslocated back to the roots via the phloem (Drew and Saker, 1984; Heuwinkel et al., 1992; Jeschke et al., 1996, 1997; Mimura et al., 1996). This favours root growth and facilitates that acquisition of additional Pi. The proteins involved in the retranslocation process are not known. To investigate Pi uptake rates and Pi translocation to the shoot, we used 33Pi as a radiotracer. These analyses revealed that, during growth in low-Pi conditions, at4 showed a small reduction in Pi uptake rate relative to wild type but, consistent with the Pi content studies, at4 plants showed a greater proportion of the 33Pi radiotracer in the shoots. Increased Pi accumulation in the shoot could result from an increase in the amount of Pi transferred to the shoot or a decrease in the amount retranslocated to the roots, or possibly a combination of both. In our experiments, Pi uptake rates were measured over a period of 60 min and, consequently, it is not possible to determine whether the translocation or retranslocation is altered. However, differences in allocation observed in plants harvested after 2 weeks of growth in low Pi versus those harvested after 4 weeks growth in low Pi provide support for an alteration in retranslocation. at4 plants harvested after only 2 weeks growth did not show an increase in the accumulation of the radiotracer in the shoots relative to wild type. It is apparent from the Pi uptake rates that the plants harvested at 2 weeks were not as Pi starved as those harvested at 4 weeks and so it is possible that the altered Pi accumulation phenotype was not revealed at this time point. In contrast, after 4 weeks of growth under low-Pi conditions, the allocation phenotype was obvious. Pi starvation is accompanied by an increase in the retranslocation activity (Jeschke et al., 1997), therefore the differences in our observations at 2 and 4 weeks may be a reflection of the degree of Pi starvation and the increase in Pi accumulation in the at4 shoots may results from a reduction in the retranslocation activity, as this activity would be most prominent in the 4-week samples relative to the 2-week samples. The location of the expression of At4 in the vascular tissue is consistent with a role in translocation or retranslocation. pho2, an Arabidopsis mutant that shows elevated Pi levels in the shoots, is predicted to play a role in Pi allocation also (Delhaize and Randall, 1995; Dong et al., 1998). However, in contrast to at4, pho2 accumulates Pi in the shoots under both high- and low-Pi growth conditions and shows higher Pi uptake rates relative to wild type. Decapitation of the plants returns the uptake rate to wild-type levels. In at4, the rate of Pi uptake into the roots is similar or slightly lower than wild type and is not altered when the shoots are removed.
Pi starvation is accompanied not only by increased root growth but also by alterations in root architecture including a reduction in growth of the primary root and subsequently an increase in the production of lateral roots (Lopez-Bucio et al., 2002; Williamson et al., 2001). Elongation of the primary root is correlated with the Pi content of the shoot and is thought to be controlled both systemically in response to the Pi content of the shoot and in response to local Pi conditions sensed by the primary root meristem (Delhaize and Randall, 1995; Dong et al., 1998; Ticconi et al., 2004; Williamson et al., 2001). During growth in high-Pi conditions, the rate of elongation of the primary root in at4 is the same as wild type. Under Pi-starvation conditions, primary root growth in the wild-type and at4 lines is reduced relative to growth rates at high Pi, but the decrease is smaller in the at4 mutant and the primary root grows at a faster rate than wild type. It seems likely that this phenotype is an indirect effect arising from the elevated levels of Pi in the shoots.
In addition to Pi-starvation-inducible activation of transcription, we suggest that At4 is regulated by a small RNA also. The highly conserved 22-nt region in the middle of all the Arabidopsis At4 family genes and conserved across the Mt4/TPSI1 gene family in broad range of plant species is suggestive of this type of regulation. We were able to detect the presence of a small RNA, of approximately 22 nt, in RNA from Pi-starved roots that is complementary to the conserved sequence present in At4. This species was not detected in RNA from plants grown in high Pi. Furthermore, the presence of this small RNA correlated with the presence of RNAs between 75 and 100 nt that were detected with an antisense probe to the conserved region. We suggest that these RNAs are possible cleavage and degradation products of the At4 transcript, because they are not present in RNA from the at4 mutant but are detected in Pi starved roots of the at4 mutant complemented with the wild-type At4 genomic fragment. In addition, the products occur in both Pi starved and Pi sufficient roots of at4 complemented with the Mt4 gene driven by the 35S promoter. A family of small RNAs, miR399(a–f), are partially complementary to the At4 conserved 22-nt sequence. The 12 nt at the 5′ end of miR399b shows a perfect match, while the 3′ half of the miR399b contains some mismatches. As complementarity at the 5′ end is considered most important (Bartel, 2004), it is possible that miR399b could regulate At4. The miR399 family is one of the largest miRNA families in plants and contains six members in Arabidopsis, nine members in rice and is also conserved in M. truncatula, although the number of members is not known for this species (Sunkar and Zhu, 2004). While the small RNA that we detect might be miR399b or a related family member, previous studies identified the miR399 in stress and ABA treated tissues (Sunkar and Zhu, 2004). Pi-starvation-induced expression has not been reported previously. Interestingly, a Pi transporter, Pht1;7, is among the list of potential targets for the miR399 RNAs (Jones-Rhoades and Bartel, 2004). Here, we were able to detect Pht1;7 transcripts in Pi-starved roots under the growth conditions in which we detect the small RNA species in both wild-type and at4 lines.
Based on these data, we suggest that At4 is regulated both at the level of transcription and post-transcriptionally by an miRNA. The conservation of the 22-nt sequence across a broad range of plant species, including M. truncatula, Medicagosativa, tomato, rice and Arabidopsis, is also consistent with its function as an miRNA target site. Regulation via the small RNA would allow the fine tuning of At4 levels and it is possible that the decrease in At4, At4.1 and At4.2 transcripts observed following ABA treatment might be mediated by an miRNA.
In summary, we propose that At4 functions in the control of Pi allocation between the shoots and roots, possibly by influencing Pi retranslocation to the roots. How this is mediated remains to be determined. Recent evidence has indicated that the mechanism of action of some short open-reading frame mRNAs (sORF-mRNAs) involves the RNA itself. The sORF-mRNA, Enod40, plays a role in the regulation of the subcellular location of an RNA binding protein (Campalans et al., 2004). At4 has a similar sORF structure and it is possible that the At4 mRNA is the biologically active product. Alternatively, it is possible that one or more of the small ORFs present in the At4 transcript might be translated to create a small peptide that would potentially be mobile in the vascular system and might signal Pi status to control translocation events. It was shown recently that two of the multiple ORFs present in the Enod40 sequence are translated and the peptides impact plant development (Guzzo et al., 2005; Rohrig et al., 2002, 2004). Other peptides that function as signals in plants include systemin (Ryan et al., 2002), DVL (Wen et al., 2004), phytosulphokines (Bahyrycz et al., 2004; Matsubayashi et al., 2004; Sakagami, 2004), CLAVATA 3 (Lenhard and Laux, 2003; Sharma et al., 2003) and the pollen S determinants (Tang et al., 2004). In support of this hypothesis, preliminary data suggest that in vitro translation of Mt4 and At4 gene families results in small peptides (HS and MJH, unpublished data). Future work will determine whether these products occur also in vivo.
Materials and methods
Plant materials and growth conditions
Wild-type A. thaliana ecotypes Wassileskija (Ws) and Columbia (Col) were obtained from the Arabidopsis Biological Research Center (Ohio State University, Columbus, OH, USA). Seeds were surface sterilized before use in all experiments as described. Plants were grown in growth rooms or chambers at 21–23°C with a light intensity of ∼100–120 μE m−2 sec−1 and a light cycle of 10-h-light/14-h-dark, unless specifically stated. For general propagation, plants were grown on MetroMix 350 (Scotts, Marysville, OH, USA) with a 16-h-light/8-h-dark cycle.
For growth in sand, seeds were germinated and grown for 14 days on 0.5 × Hoagland agar plates (Arnon and Hoagland, 1940) containing 1 mm Pi and then transferred to pots (eight plants per pot) containing acid-washed sand. The plants were then fertilized weekly with the nutrient solutions containing different concentrations of Pi as indicated. For the P, nitrogen and sulphur deprivation experiment (Figure 1), plants were grown in sand and fertilized with 0.5 × Hoagland solution containing 1 mm Pi for 2 weeks, then subjected to fertilization regimes lacking P, nitrogen or sulphur for an additional 2 weeks. For the nitrogen-minus 0.5 × Hoagland solution (−N fertilizer), Ca(NO3)2 and KNO3 were replaced by the equimolar amount of CaCl2 and K2SO4, respectively. For sulphur-minus solution (−S fertilizer), MgSO4 was replaced by MgCl2. Experiments with defined Pi agar media were based on 0.5 × Hoagland solution with 0.8% (w/v) agar and the Pi concentration was varied as described. For the experiment shown in Table 2, the plates contained a cellophane overlay to enable easy removal of the plants.
For the hormone treatment experiments, approximately 100 surface-sterilized Arabidopsis seeds were germinated and grown hydroponically in a Magenta® GA-7 box (Sigma-Aldrich, St Louis, MO, USA) with a vented lid containing 100 ml of 0.5 × MS salts and 1% sucrose solution, pH 6.0. The plants were then transferred to 0.5 × MS without Pi and grown for a further 2 weeks to induce starvation. Plant growth regulators were added into the growth media at the end of 2-week starvation as follows: 50 μm ABA, 50 μm ACC, 100 μm SA, 10 μm IAA, 10 μm kinetin, 100 μm GA3, or 50 μm AVG. Plants were harvested 24 h later.
The analysis of the Pi transporters (Figure 4) was undertaken on plants grown in hydroponic culture. Plants were grown in 0.5 × Hoagland solution with 1 mm Pi for 2 weeks. After 2 weeks, the media was replaced with media containing 0 mm Pi and the plants were grown for an additional 2 weeks before harvest.
The analysis of the small RNA was undertaken on plants grown in sand (Figure 5) and in hydroponic culture (Figure 6). For the hydroponic experiments, approximately 200 seeds were germinated and grown in Magenta GA-7 boxes with vented lids containing 100 ml of 0.5 × Hoagland solution with 2 mm Pi for 2 weeks, with a weekly change of solution. After 2 weeks, the media was replaced either with the same Pi concentration (2 mm) for high-Pi condition or 0 mm Pi concentration for the low-Pi condition and the plants were grown for an additional 2 weeks. The media was replaced weekly.
Cloning At4.1 and At4.2
RT-PCR was used to amplify cDNA sequences containing the conserved 21-nt sequence (At4 21 nt; 5′-GGGCAACTTCGATCCTTTGGC-3′). Total RNA was isolated from roots of Arabidopsis grown in sand under Pi-starvation conditions and used as template for the synthesis of first-strand cDNA in a reverse transcription reaction using Superscript-II RT (Invitrogen, Carlsbad, CA, USA). PCR was performed with the At4 21 nt as a 5′-gene-specific primer and oligo(dT)18 as a 3′ primer. A slightly diffuse band of PCR at ∼250 bp was confirmed by agarose gel electrophoresis. The band was gel-eluted and ligated into a TA vector (Invitrogen) for Escherichia coli transformation. We isolated plasmids from 12 E. coli transformants for DNA sequencing. We found two novel partial cDNA sequences. These were used as probes on Northern blots to estimate the sizes of the full-length transcripts. Full length sequences were obtained by additional PCR-based strategies and named At4.1 and At4.2. The gene-specific primers used to amplify the longer transcripts of At4.1 and At4.2 genes are as follows: At4.1 (5′ primer for At4.1, 5′-GACTCAGGTCAGCTTAGTTAATGGTTCA-3′); At4.2 (5′ primer for At4.2, 5′-GCATATCCCTTGGAAGATTCCCTATTTC-3′).
RNA expression analyses
Total RNA was isolated using Trizol reagent. Northern blots were carried out following standard procedures. Low molecular weight, small RNA was isolated from total RNA by anion-exchange chromatography (RNA/DNA Midi Kit; Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. For RNA blot, small RNA was resolved on denaturing 15% polyacrylamide gels (Invitrogen) containing 7 m urea in Tris-Borate-EDTA (TBE) buffer, electroblotted to GT membrane (Bio-Rad, Hercules, CA, USA) using a transblot semidry transfer cell (Bio-Rad), and UV cross-linked.
DNA oligonucleotide probes were end labelled with γ-32P-ATP using T4 polynucleotide kinase. Blots were pre-hybridized for 1 h and hybridized for overnight at 37°C by using PerfectHyb plus buffer (Sigma-Aldrich). After hybridization, the blots were rinsed briefly with 4 × saline-sodium citrate (SSC) and washed with 2 × SSC at 42°C for 20 min.
RNA size markers were synthesized from in vitro transcription reaction. Briefly, the pBluescriptII-KS vector linearized by digestion with restriction enzymes, including KpnI, XhoI, PstI and SacI, was used as template for in vitro transcription by using the T7 Maxiscript kit (Ambion, Austin, TX, USA) according to the manufacturer's protocol. RNA was radiolabelled with α-32P-CTP to enable autoradiographic detection.
Screening and Identification of the At4 mutant
A PCR-based method was used to screen the Wisconsin T-DNA lines (a population of 60 480 T-DNA insertion lines available through the Arabidopsis Knockout Facility at the University of Wisconsin, Madison, WI, USA) to isolate a line with a T-DNA insertion in the At4 gene. The flanking sequences of the T-DNA insertion in the At4 gene were confirmed by genomic PCR with primers as follows: a T-DNA left border primer, JL202 (5′-CATTTTATAATAACGCTGCGGACATCTAC-3′); a 5′ primer for At4, 4a (5′-ACAAAGAGAGAAGCCATAAAAACCCTAAC-3′); a 3′ primer for At4, 4b (5′-ATAGGAACACACCTGAATGGTGCATCACA-3′).
Analysis of primary root growth
Plants were germinated and grown vertically on 0.5 × Hoagland agar plates containing no additional Pi or 500 μm Pi. Root growth was measured from scanned images of plants grown for 9, 11 and 14 days after planting. Mean growth rates were calculated as mm day−1 from at least 15 seedlings.
An initial uptake experiment was carried out with pools of 50 plants grown on nylon meshes floating on 0.5 × Hoagland media [0.3% (w/v) agar] containing either 2 or 500 μm Pi. Following 2 weeks of growth, the plants were transferred to Pi uptake solutions containing either 2 or 500 μm Pi supplemented with P33 as a radiotracer and Pi uptake was measured (Shin et al., 2004).
To analyze uptake and distribution within the plant, sterilized seeds were germinated and grown on cellophane supported on 0.5 × Hoagland, 5 μm Pi agar plates for 2 or 4 weeks. Seedlings were carefully removed from the plates and rinsed with nutrient solution, then placed on the surface of microscopic slides that were positioned at a slanted angle in Petri dishes in Pi-uptake solution. The Pi-uptake solution was 0.5 × Hoagland, 5 μm Pi with Pi33 at 1 μCi ml−1. Seedlings were incubated in uptake solution for up to 1 h, rinsed briefly in the ice-cold nutrient solution for ∼2–3 min, gently blot-dried on the 3 m filter paper (Whatman, Florham Park, NJ, USA) and the shoots were separated from the roots for individual scintillation counting for autoradiography.
Pi content determination
Tissues were rinsed with distilled water and gently blot-dried on filter paper. The tissues were weighed immediately for fresh weights, frozen in liquid nitrogen and ground to a fine powder, and suspended in 1% acetic acid. Pi content of the tissue suspension was measured using a phosphomolybdate colorimetric assay (Ames, 1966).
Constructions of recombinant vectors
To create the At4 promoter-Gus fusion, the 2-kb region upstream from the second putative ATG of the At4 cDNA sequence (GeneBank ID: AY040018) was amplified by PCR using the following primers At4fwd (5′-forward primer, 5′-AACTGAATTCCTCGAACTATTTGCGTATTGTTC-3′) and At4revA (3′-reverse primer, 5′-ATGGCCATGGTTTGTGTGTTTGGTGTTGTGCCA-3′), which contain EcoRI and NcoI restriction sites, respectively. The PCR product was digested with EcoRI and NcoI, and cloned into pCambia3301 upstream of the GUS gene creating a translational fusion.
For construction of 35S promoter–Mt4 gene, the 0.9-kb genomic region downstream from the transcription initiation site of the Mt4 gene was amplified by PCR using primers Mt4fwd (5′-forward primer, GATCGAATTCCAACCTACTTAAGTCATC) and Mt4rev (3′-reverse primer, CTAGGGATCCGTCGACTCGATCCATATCC) containing EcoRI and BamHI restriction sites, respectively. The amplified fragment was subcloned into pRTL2 vector using restriction sites EcoRI and BamHI, generating a new recombinant plasmid pRTL2-35SProm-Mt4-35Sterminator. The 35Spromoter-Mt4-35Sterminator sequence was digested out by HindIII restriction enzyme and cloned into a binary vector pSKI006 linearized with HindIII.
For complementation of At4 with the wild-type At4 gene, the genomic fragment encompassing the At4 gene sequence was obtained by PCR using genomic DNA isolated from wild type. The primers were designed to amplify 3197 nt of genomic fragment that contains approximately 2 kb of At4 promoter region. The primers were At4fwd as used above and At4revB (3′-reverse primer, 5′-GAACGGATCCTGGAACTCAAACTAGAGGTC-3′) containing EcoRI and BamHI sites, respectively. The amplified product digested with EcoRI and BamHI was cloned into pCambia3300 binary vector using the same restriction sites.
The recombinant binary vectors based on pSKI006 and pCambia3300/3301 were used to transform Agrobacterium tumefaciens ASE and C58C1 strains by using a freeze–thaw method, respectively. Arabidopsis transformation was performed by using a flower-dipping method as described (Clough and Bent, 1998).
The At4 mutant was crossed to the double Pi transporter knockout mutant (pht1;1Δ4Δ), which has insertions in the Pht1;1 and Pht1;4 Pi transporter genes. Plant lines carrying T-DNA insertions in all three genes (At4, pht1;1Δ4Δ) were identified from F2 progeny by PCR.
GUS staining was performed according to the general protocol as described. For histochemical localization of GUS activity, the stained tissues were subjected to tissue fixation and procedures for sectioning as described (Takechi et al., 1999). The fixed tissues were embedded in Technovit resin (Energy Beam Sciences, East Granby, CT, USA) for sectioning (10 μm) with a rotary microtome.
The authors thank members of the M.J. Harrison and R. Chen laboratories for helpful discussions. This work was supported by the Samuel Roberts Noble Foundation and by Atlantic Philanthropies Inc.