Present address: Instituto de Recursos Naturales y Agrobiología de Salamanca, Consejo Superior de Investigaciones Científicas, 37008 Salamanca, Spain.
Genome-wide reprogramming of metabolism and regulatory networks of Arabidopsis in response to phosphorus
Article first published online: 29 NOV 2006
Plant, Cell & Environment
Volume 30, Issue 1, pages 85–112, January 2007
How to Cite
MORCUENDE, R., BARI, R., GIBON, Y., ZHENG, W., PANT, B. D., BLÄSING, O., USADEL, B., CZECHOWSKI, T., UDVARDI, M. K., STITT, M. and SCHEIBLE, W.-R. (2007), Genome-wide reprogramming of metabolism and regulatory networks of Arabidopsis in response to phosphorus. Plant, Cell & Environment, 30: 85–112. doi: 10.1111/j.1365-3040.2006.01608.x
- Issue published online: 29 NOV 2006
- Article first published online: 29 NOV 2006
- Received 14 July 2006; received in revised form 26 September 2006; accepted for publication 2 October 2006
- expression profiling;
- transcription factors
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
Affymetrix ATH1 arrays, large-scale real-time reverse transcription PCR of ∼ 2200 transcription factor genes and other gene families, and analyses of metabolites and enzyme activities were used to investigate the response of Arabidopsis to phosphate (Pi) deprivation and re-supply. Transcript data were analysed with MapMan software to identify coordinated, system-wide changes in metabolism and other cellular processes. Phosphorus (P) deprivation led to induction or repression of > 1000 genes involved in many processes. A subset, including the induction of genes involved in P uptake, the mobilization of organic Pi, the conversion of phosphorylated glycolytic intermediates to carbohydrates and organic acids, the replacement of P-containing phospholipids with galactolipids and the repression of genes involved in nucleotide/nucleic acid synthesis, was reversed within 3 h after Pi re-supply. Analyses of 22 enzyme activities revealed that changes in transcript levels often, but not always, led to changes in the activities of the encoded enzymes in P-deprived plants. Analyses of metabolites confirmed that P deprivation leads to a shift towards the accumulation of carbohydrates, organic acids and amino acids, and that Pi re-supply leads to use of the latter. P-deprived plants also showed large changes in the expression of many genes involved in, for example, secondary metabolism and photosynthesis. These changes were not reversed rapidly upon Pi re-supply and were probably secondary in origin. Differentially expressed and highly P-specific putative regulator genes were identified that presumably play central roles in coordinating the complex responses of plants to changes in P nutrition. The specific responses to Pi differ markedly from those found for nitrate, whereas the long-term responses during P and N deprivation share common and non-specific features.
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
Phosphorus (P) is an essential component of intermediates in central and energy metabolism, signalling molecules and structural macromolecules like nucleic acids and phospholipids. Plants obtain P as inorganic phosphate (Pi). Pi often limits plant growth because the levels in the soil are often low; much is covalently or non-covalently bound, and mobility is poor (Marschner 1995). Pi enters metabolism via non-dedicated pathways. By far, the most important route is via synthesis of ATP from adenosine diphosphate (ADP) and Pi, and the ensuing transfer to other nucleotides and phosphorylated intermediates. While most of the Pi is rapidly recycled, some is transferred into P-containing end products. This contrasts with inorganic nitrogen (N) and sulphur (S), which are assimilated by clearly demarcated assimilatory pathways that are regulated in response to the balance between macronutrient availability and utilization in metabolism and growth (Kopriva & Rennenberg 2004; Saito 2004; Krapp, Saliba-Colombani & Daniel-Vedele 2005). There are important unanswered questions about how central carbon and energy metabolism are coordinated with the availability and utilization of Pi. P, like N and S, is recycled by degrading intermediates, cofactors and structural macromolecules. Excess Pi, like nitrate (NO3-) and sulphate (SO42−), is mainly stored in the vacuole.
Pre-genomic studies provide insights into mechanisms that control Pi uptake and utilization in plants (Abel, Ticconi & Delatorre 2002). Physiological responses in Pi-deficient plants include an increased root/shoot ratio (Lynch 1995), alternations in root architecture (Williamson et al. 2001; Lopez-Bucio et al. 2002), production of lateral roots in Pi-rich patches in the soil (Robinson 1994) and the proliferation of long root hairs (Bates & Lynch 2001). When Pi is limiting, Pi uptake is promoted by increased expression of Pi transporters (Mudge et al. 2002; Smith et al. 2003), synthesis and secretion of organic acids to solubilize Pi in the soil and apoplast (Raghothama 1999), and induction of phosphatases (del Pozo et al. 1999; Li et al. 2002) andRNAases (Bariola et al. 1994) that mobilize organically bound P inside and outside the plant. Changes in metabolism allow remobilization of Pi in the plant, including a general decrease in the levels of P-containing intermediates and cofactors such as nucleotides (Zrenner et al. 2006), replacement of phospholipids by sulpholipids and galactolipids (Dörmann & Benning 2002; Kelly, Froehlich & Dörmann 2003) and a general decrease in the level of RNA (Hewitt et al. 2005). Vacuolar Pi is remobilized, although the mechanisms are not yet elucidated. The existence of mutants with altered allocation of Pi between the shoot and root (see subsequent discussion) implies that plants regulate the long-distance transport and allocation of Pi. Further changes in response to P limitation include the accumulation of carbohydrates and secondary metabolites like anthocyanins.
Pi sensing and signalling have been extensively characterized in bacteria and yeast (Wanner 1993; O'Neill et al. 1996). Less is known about Pi signalling in plants (Abel et al. 2002). Several mutants have been identified with altered levels or distribution of Pi (pho1, pho2 and pup1; see Poirier et al. 1991; Delhaize & Randall 1995; Trull & Deikman 1998). Molecular cloning reveals that PHO1 encodes the first member of a novel EXS domain (PFAM entry 03124; http://www.sanger.ac.uk) transporter family (Hamburger et al. 2002). It is probably involved in Pi transport into the xylem (Wang et al. 2004) rather than signalling. Cloning of PHO2 shows that it encodes an E2 conjugase, implicating Pi signalling with the targeted protein degradation pathway (Aung et al. 2006; Bari et al. 2006). Interestingly, PHO2 is the target of Pi-regulated microRNA399, which has been proposed as a long-distance signal to adjust root Pi uptake and translocation to shoot Pi status (Bari et al. 2006).
Mutants that are impaired in aspects of the Pi-stress response have also been identified (pho3, psr1, pdr2 and phr1; see Chen et al. 2000; Rubio et al. 2001; Zakhleniuk, Raines & Lloyd 2001; Ticconi et al. 2004). PHO3 has recently been shown to encode the sucrose transporter SUC2 (Lloyd & Zakhleniuk 2004) indicating that it acts rather indirectly. The global changes in gene expression in PHO3 have no similarity with those in P-deficient plants but are more closely related to those resulting from C and N deficiencies (Usadel et al. 2005).
A clear role has been established for PHR1 in P signalling. This MYB-like transcription factor has homology to PSR1, which is involved in Pi sensing in Chlamydomonas reinhardtii (Wykoff et al. 1999), and regulates the expression of target genes like acid phosphatase (AtACP5), AtIPS1, PHT1.1 and RNS1 (Martin et al. 2000; Rubio et al. 2001), and other Pi starvation-induced genes including microRNA399 genes (Bari et al. 2006) by binding to an imperfect palindromic 8-bp sequence (GNATATNC) in their promoter (Rubio et al. 2001; Franco-Zorilla et al. 2004; Hammond, Broadley & White 2004).
Recent studies of T-DNA insertion mutants for SUMO E3 ligase SIZ1 also implicate protein sumoylation in Pi responses (Miura et al. 2005). In P-replete conditions, Pi levels and transcript levels of a subset of P-responsive genes (AtPHT1.4, AtPS2 and AtPS3) are higher in siz1 mutants than in wild-type plants. In low Pi, siz1 shows exaggerated phenotypic changes reminiscent of Pi deficiency including increased lateral root and root hair growth, and increased anthocyanin even though the internal Pi content resembles the wild type, and the induction of some low Pi-induced genes (AtIPS1, AtRNS1) is slower than in wild-type plants. Interestingly, PHR1, which regulates these two genes (see previous discussion) contains sumoylation sites. These results indicate a complex role for sumoylation in Pi responses, influencing events upstream from PHR1 and acting positively and negatively on different sectors of Pi-signalling pathways.
Expression profiling studies with a 6K Arabidopsis expressed sequence tag (EST) array (Wu et al. 2003), an 8K Arabidopsis GeneChip (Affymetrix, Santa Clara, CA, USA; Hammond et al. 2003) and a 9K rice microarray (Wasaki et al. 2003) have shown that P deprivation modifies the expression of numerous genes. The changes include down-regulation of genes for photosynthesis and N assimilation, alterations in the balance between synthetic and degradative carbon metabolism, down-regulation of protein synthesis and up-regulation of proteases, and changes of several transcription factors and other signalling-related genes. Recently the short-, mid- and long-term responses of Arabidopsis plants to P deprivation have also been studied using 22K ATH1 GeneChips (Misson et al. 2005). In all these studies, however, changes during P deprivation have been documented without further analysis to distinguish between direct responses to changing Pi and indirect effects due to slow growth, carbohydrate accumulation or other secondary effects. ATH1 data are also available in the public domain (NASCArrays database, http://affymetrix.arabidopsis.info/) for a comparison of pho1 mutant and wild-type leaves. The pho1 mutant is unable to transport Pi from roots to shoots, leading to Pi starvation in above-ground tissues (Poirier et al. 1991).
Expression arrays have inherent limitations when employed on their own. Genes represented by unspecific probe sets and with low expression levels cannot be reliably measured (Redman et al. 2004; Usadel et al. 2005). Further, changes of transcript levels do not necessarily show that the levels of the encoded proteins have changed, or the relevant metabolic pathway or biological process has been affected. The work presented here extends previous analyses of Pi responses in two ways. Firstly, we performed a multilevel analysis, including commercial 22K ATH1 arrays, real-time reverse transcriptase (RT)-PCR profiling for ∼ 2200 putative transcription factor genes (Czechowski et al. 2004) and other important gene families involved in the response to P, and combined these with robot-based measurements of enzyme activities (Gibon et al. 2004) and metabolite analyses. Secondly, to allow P-specific responses to be identified, we tested whether the changes in Pi starvation were rapidly reversed when Pi was re-supplied, and compared the response to Pi to that of N depletion and re-addition (Scheible et al. 2004) and to other stimuli. The aims were to (1) identify which genes are specifically regulated by changes in Pi, (2) investigate whether changes in transcripts lead to changes in enzyme activities and metabolism and (3) identify a robust set of candidate genes that could be involved in Pi signalling.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
Plant growth media and conditions
The sterile full nutrition (FN, +P, 3.0 mM Pi) media were described previously (Scheible et al. 2004). The ‘reduced P’ medium contained 0.2 mM KH2PO4/K2HPO4[potassium phosphate (KPi), pH 5.8], 2.8 mM KCl, and all other components as in FN media. No P (−P) medium did not contain any KPi and otherwise all components as reduced P medium. Wild-type Columbia-0 Arabidopsis seedlings (120–150 per flask) were grown in axenic culture under constant low light as described in Scheible et al. (2004) in FN or reduced P medium. After 7 d, the media were changed in all flasks. The seedlings growing in FN media received fresh FN media. The seedlings growing in reduced P media did not receive P media (−P). +P media were renewed on day 8 in FN cultures to prevent N limitation, because by this stage, N was rapidly depleted by growing seedlings in FN culture (cf. Scheible et al. 2004).
Pi addition and seedling harvest
On day 9, FN cultures and some of the −P cultures were harvested. At the same time, all the other flasks of P-starved cultures were opened and either re-closed without addition or after adding 150 µL 100 mM KPi (pH 5.8). The added liquid was allowed to disperse without changing the shaking speed. Groups of −P flasks that received KPi or no addition were harvested after 12, 30 and 75 min, and after 3, 8 and 24 h. Whole seedlings were harvested as described (Scheible et al. 2004).
Extraction and assay of enzyme activities
RNA preparation and ATH1 array hybridization
Three biological replicates (i.e. the contents from three flasks) from the same experiment were pooled to prepare total RNA. The preparation and quality control of RNA and biotin labelling of the cRNA target were as described by Czechowski et al. (2004). Hybridization, washing, staining and scanning procedures were performed at RZPD (Berlin, Germany) as described in the Affymetrix technical manual (Affymetrix, Santa Clara, CA, USA).
ATH1 data analysis and MapMan display
The raw Affymetrix signals (CEL files) were processed using log-scale robust multiarray analysis (RMA) open access software (http://stat-www.berkeley.edu/users/bolstad/RMAExpress/RMAExpress.html). It is based on the quantile normalization method and has better precision than MicroArray Suite 5.0 (Affymetrix) and dCHIP (http://www.dchip.org/), especially for low expression values (Irizarry et al. 2003). RMA-calculated signal intensities for all 22 750 ATH1 probe sets with MAS5 ‘present’ or ‘absent’ calls from all nine arrays are compiled in Supplementary Table S1.
The averaged signals for a given treatment (three biological replicates for full nutrients, two biological replicates for 30 min 0.5 mM KPi addition, two biological replicates for 3 h 0.5 mM KPi addition) were expressed relative to those in P-deficient seedlings (two biological replicates for no addition), converted to a log 2 scale and imported into the MapMan software, which converts the data values to a false colour scale and paints them out onto the diagrams (Thimm et al. 2004). A downloadable version for local application and a servlet version are available at http://gabi.rzpd.de/projects/MapMan/. The web site contains instructions for the installment and use of the software. The downloadable installers include (1) the Affymetrix experimental data sets presented in the paper, (2) a selection of schematic maps of metabolism and cellular processes and (3) mapping files that structure the Arabidopsis genes represented on the ATH1 array into BINS and sub-BINS for display on the schematic maps of metabolism and cellular processes. The overview figures in this paper are prepared using version 190705. In addition, functional categories of the MapMan annotation (Thimm et al. 2004; Usadel et al. 2005) were tested for significance of expression change by applying a two-sided Wilcoxon rank sum test, performed in the statistical software environment R (R Development Core Team 2005). The log 2 ratios for all ‘present’-called genes in a particular MapMan BIN were compared to the ratios of all remaining genes that were called ‘present’ on the chips.
Statistical analysis of ATH1 data
The data from Supplementary Table S1 were split into three sets. These were analysed separately using the LIMMA (v2.4.13) bioconductor open access software package (by reading the data and creating an expression set of the data after log 2 transformation). A linear model was fitted and contrasts of ‘FN versus P starved’, ‘re-addition 30 min versus P starved’ as well as ‘re-addition 3 h versus P starved’ were extracted. For all data sets, an empirical Bayes approach was used to compute moderated t-statistics (Smyth 2004), and false discovery rate (FDR) correction (Benjamini & Hochberg 1995) was applied to account for testing of multiple genes.
Real-time RT-PCR analysis
Sequences of RT-PCR primers for transcription factor genes, real-time PCR conditions, data analysis and procedures for cDNA synthesis were as in Czechowski et al. (2004, 2005).
RESULTS AND DISCUSSION
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
Arabidopsis seedlings were grown in sterile liquid culture under continuous light and low levels of sucrose in the medium (Scheible et al. 2004) to minimize diurnal changes in metabolism that might complicate interpretation of the results (Scheible, Krapp & Stitt 2000; Matt et al. 2001a,b). FN control cultures (+P) were grown with 3 mM KPi. To obtain Pi-deficient seedlings, the cultures were grown for 7 d with 0.2 mM Pi, which was sufficient for maximal biomass accumulation, but did not lead to accumulation of Pi (data not shown), and was then deprived of exogenous Pi for 2 d (−P). P-replete (+P) seedlings and some batches of P-deprived (−P) seedlings were harvested after 9 d. Other batches of −P seedlings received 0.5 mM KPi and were incubated for up to 24 h before harvest. This experimental design allowed two criteria to be applied to identify P-responsive genes: they should (1) show marked changes of transcript levels between P-replete and P-deprived seedlings that are (2) partly or totally reversed after short-term re-supply of Pi.
Physiological and metabolic responses to P deprivation and re-supply
By day 9, all seedlings (+P and −P) had developed cotyledons and first leaves (Fig. 1). P-deprived seedlings exhibited typical stress phenology (see Introduction) including dark green leaves and pronounced root and root hair growth. Triplicate samples were taken during the first 24 h after re-supplying Pi to investigate the levels of central metabolites. All metabolite data are available in the supplementary material (Supplementary Table S2); selected examples are shown in Fig. 2a–e, and the remainder is summarized in Fig. 2f. Samples from the −P and +P treatments were measured at the beginning and end of the 24 h period.
Internal Pi was ∼ 100-fold lower in P-deficient than P-replete seedlings. Internal Pi increased significantly within 75 min of Pi re-supply, was ∼ 4-fold increased after 8 h, and then declined slightly after 24 h (Fig. 2a). This transient increase of Pi in the seedlings is consistent with rapid uptake and utilization of the added Pi. Adenine nucleotides were twofold lower in P-deficient seedlings (Fig. 2b,c). After Pi re-supply, ATP rose promptly, whereas ADP remained unaltered at first. This is consistent with a stimulation of ATP synthesis as Pi becomes available. Because ATP turns over rapidly, the Pi incorporated will be recycled or transferred to other metabolites. By 3 h, ADP, ATP and uridine diphosphate-glucose (UDP-Glc) (Fig. 2e) were rising. The latter represents the major part of the uridine nucleotide pool (Dancer, Neuhaus & Stitt 1990).
Glc6P (Fig. 2d), Fru6P and Glc1P (Fig. 2f) were very low in P-deficient seedlings, as expected (Hurry, Furbank & Stitt 2000). They increased twofold within 12 min of Pi addition, indicating that Pi is rapidly taken up and assimilated. Metabolites further down in glycolysis [e.g. glycerate-3-phosphate, glycerate-2-phosphate, phosphoenolpyruvate (PEP)] increased in P-deprived seedlings. This did not represent a major sequestration of Pi because absolute levels were much lower than those of the hexose phosphates and nucleotides (Supplementary Table S2). These metabolites, especially PEP, declined when Pi was re-supplied, indicating that the use of glycolytic intermediates was activated after re-supply of Pi.
Pi-deficient seedlings showed a marked accumulation of starch, sucrose and, to a lesser extent, reducing sugars (Fig. 2f). There was a general increase of organic acids; citrate rose almost 10-fold, and there were also marked increases of fumarate, malate and oxoglutarate (Fig. 2f). Organic acids and carbohydrates did not fall markedly in the first 24 h after Pi re-addition.
Pi-deficient seedlings had unaltered or slightly increased levels of most major amino acids, especially glycine, and elevated levels of several minor amino acids including the aromatic amino acids and His, Arg and Thr (Fig. 2f). Methionine was lower in P-deficient seedlings, indicating an inhibition of sulphur metabolism. After P re-supply, amino acids remained unaltered for the first 3 h. There was a general decrease of amino acids from 8 h onwards; the level in some cases (e.g. Gly, Ser, His, Phe, Tyr) remained above, and the level in others (Gln, Glu, Asp, Ala, Asn, Ile, Leu, Val, Lys, Met) fell below that of +P seedlings.
Except for hexose phosphates and ATP, the changes of metabolites in P-deprived seedlings were not reversed in the first 3 h after Pi re-supply. Transcript profiling was performed 30 min and 3 h after Pi re-supply to reveal changes that occur independently of changes of major metabolites like carbohydrates, organic acids or amino acids.
Expression profiling of Arabidopsis seedlings with Affymetrix ATH1 GeneChips
Two independent experiments were carried out at an interval of 6 months to harvest samples for transcript profiling from entire +P seedlings, −P seedlings and −P seedlings 30 min and 3 h after Pi re-supply. Materials for each sample were pooled from two separate flasks, each containing ∼ 120 seedlings. ATH1 data from a third biological replicate of +P (i.e. FN seedlings) (Scheible et al. 2004) were also taken into account for statistics. Genes/ATH1 probe sets showing (1) a ≥ 2-fold change between −P and +P conditions in each replicate and (2) an adjusted P-value < 0.05 (Benjamini & Hochberg 1995) were identified from the seedling experiments. Application of these criteria indicated that 570 ATH1 probe sets showed an increased signal and 488 showed a decreased signal in Pi deprivation (Fig. 3a).
There were few changes in transcript levels 30 min or 3 h after re-supply of Pi, apart from the reversal of the changes of a subset of the genes that were induced in P deprivation (Supplementary Table S1; Supplementary Fig. S1, panel g). The signal intensities of some of the strongly (> 5-fold) P deprivation-induced genes were slightly reversed within 30 min, and most were strongly (> 90%) reversed after 3 h Pi re-supply (Fig. 3b, Supplementary Table S1), indicating that these genes are involved in direct responses to P availability. Of the less strongly P deprivation-induced genes (two- to fivefold), about 40–50% showed a marked (> 50%) reversion in the first 3 h after Pi re-supply (Fig. 3a). Only a few of the genes that were repressed by P deprivation showed a marked reversion (i.e. an increase in transcript abundance after 3 h Pi re-supply). There were virtually no genes that showed significant transient expression changes after Pi re-addition. In contrast, transient changes were frequently found after re-supply of nitrate or sugar to N- or C-deprived seedlings (Scheible et al. 2004; Osuna et al. in press).
Table 1 shows the annotations and known stimuli for all genes/probe sets with > 20-fold changed mean expression/signal intensity plus several additional genes that showed more than 10-fold altered signal intensity (cf. Supplementary Table S1). Among these genes are (1) several that are known to be induced during P deprivation (e.g. galactolipid synthesis, Pi transport, acid phosphatase, ribonuclease; see Introduction and subsequent discussion), (2) a few whose annotations suggest functions in Pi signalling (e.g. MAP3K, transcription factors, SPX (PFAM entry PF03105; http://www.sanger.ac.uk) domain protein), and (3) some very highly induced genes that do not have clear functions attributed (e.g. the segmental duplicated genes At1g17710/At1g73010 or the homologous genes At5g20790/At3g43110). Cross comparisons with data from AtGenExpress and our own published (Scheible et al. 2004) and unpublished data reveal that many of these highly Pi-responsive genes are also transcriptionally induced by salt, cold, osmotic stress or abscisic acid, or during senescence or seed development (Table 1). This may indicate that some of these responses include a re-adjustment of Pi utilization. It is already known that low temperature leads to acute Pi limitation (Hurry et al. 2000; Stitt & Hurry 2002), and senescence and seed fill require mobilization and storage of Pi. Several P stress-induced genes were not significantly influenced by any of the > 30 stimuli investigated, and thus appear to be highly P specific.
|Rank||Probe set||Arabidopsis Gene Identifier (AGI)||Annotation||Other stimuli|
|1||262369_at||At1g73010||Expressed protein, putative phosphatase||OS|
|3||259399_at||At1g17710||Expressed protein, putative phosphatase||None found|
|4||266070_at||At2g18660||Expansin protein||ST UVB –N|
|5||259221_s_at||At3g03530 At3g03540||Phospholipases C2||OS|
|7||266132_at||At2g45130||SPX domain-containing protein||None found|
|8||260097_at||At1g73220||Sugar/phosphate/anion transporter||OS ABA SEED|
|9||246071_at||At5g20150||SPX domain-containing protein||OS CO|
|10||266267_at||At2g29460||Putative glutathione S-transferase||ST OS UVB –N BIO SEN SEED|
|12||245038_at||At2g26560||Lipid acyl hydrolase (PLP2)||ST OS UVB|
|13||264400_at||At1g61800||Glucose-6-phosphate/phosphate translocator (GPT2)||OS CO UVB –N|
|14||266766_at||At2g46880||Calcineurin-like phosphoesterase||None found|
|15||266184_s_at||At2g38940 At3g54700||Phosphate transporters (Pht1;4/Pht1;7)||CO|
|16||258856_at||At3g02040||Glycerophosphoryldiester phosphodiesterase (SRG3)||CO SEN|
|18||245531_at||At4g15100||Serine carboxypeptidase S10||ST|
|20||260140_at||At1g66390||MYB transcription factor (PAP2)||OS –N –S|
|21||266743_at||At2g02990||Ribonuclease (RNS1)||OS ABA WND SEED|
|22||264893_at||At1g23140||C2 domain-containing protein||OS CO|
|23||262229_at||At1g68620||Expressed protein||ST OS ABA SEN|
|24||252984_at||At4g37990||Aromatic alcohol:NADP+ oxidoreductase (ELI3-2)||ST BIO|
|25||258887_at||At3g05630||Phospholipase D p2||OS ABA SEN SEED|
|26||267425_at||At2g34810||Berberine-bridge-forming enzyme||OS ABA ZEA|
|27||246075_at||At5g20410||Monogalactosyldiacylglycerol synthase||None found|
|28||251143_at||At5g01220||Sulpholipid synthase (SQD2)||None found|
|31||250083_at||At5g17220||Glutathione S-transferase||–N SEN|
|35||252414_at||At3g47420||Glycerol-3-phosphate transporter||CO SEN|
|36||260405_at||At1g69930||Glutathione S-transferase||ST OS BIO SEN|
|38||263210_at||At1g10585||bHLH transcription factor||ABA WND|
|47||258158_at||At3g17790||PAP17, ACP5||ST OS SEN|
|52||254767_s_at||At4g13290||Cytochrome P450 CYP71A19||–C|
|53||261215_at||At1g32970||Subtilase family protein||ST|
|54||258570_at||At3g04530||Phosphoenolpyruvate carboxylase kinase 2 (PPCK2)||NO3|
|59||264783_at||At1g08650||Phosphoenolpyruvate carboxylase kinase 1 (PPCK1)||ST OS CO|
The global Pi response of genes assigned to different functional categories, and comparison to the global N response
P and N deprivation lead to several shared long-term whole plant phenotypes. For example, both reduce whole plant growth, lead to preferential root growth and a decreased shoot : root ratio, changes in root architecture, early flowering and senescence, as well as reduced photosynthetic capacity, starch accumulation and anthocyanin production. Comparison of the molecular effects of both nutrient deprivations should therefore reveal the genes that are specific for P or N, or common to both.
Figure 4 compares the global responses to P and N. The genes represented on the ATH1 arrays were assigned to > 700 functional categories using the MapMan hierarchical ontology (http://gabi.rzpd.de/projects/MapMan). Wilcoxon non-parametric test was then used to calculate the probability that the response of the genes assigned to a functional category (BIN or sub-BIN) was statistically different from the response of all the other genes on the array. The results for the upper-level categories and subcategories that showed statistically significant changes (P < 10−4) are displayed on a monochromic scale (the P-values for all the BINS and sub-BINS are listed in Supplementary Table S3). The average change of the signal for the genes in each functional category is also shown, using a false colour scale (white indicates no change, and increasingly deep blue and red signify an increasingly large increase and decrease). The analysis was performed for two comparisons: FN versus starved seedlings, and 3 h re-supply versus starved seedlings, using nutrient-deficient seedlings as the reference.
Three general trends emerge. Firstly, the responses to P and N deprivation share many common features (lanes 2 and 4). Both lead to the repression of many genes involved in photosynthesis, tetrapyrrole synthesis, nucleotide metabolism, amino acid activation, protein synthesis and protein folding, and the induction of many genes involved in phenylpropanoid and flavonoid metabolism, stress and categories involved in signalling including transcription factors, protein kinases and phosphatases, receptor kinases and calcium-binding proteins. Secondly, there are marked differences between the response in nutrient deficiency and the response after re-supplying the nutrient. For example, re-supply of P and N (lanes 1 and 3) does not reverse the repression of genes assigned to photosynthesis. Both also induce genes assigned to nucleotide metabolism, even though these are not strongly repressed in deprived seedlings. Thirdly, the differences between P and N are especially clear after nutrient re-supply. For example, re-supply of P leads to repression, and N addition leads to induction of genes encoding enzymes in glycolysis including PEPCase and pyruvate kinase. P re-supply, but not N re-supply, represses genes involved in lipid degradation and glycolipid synthesis, genes involved in ethylene, jasmonic acid and salicyclic acid metabolism, genes encoding acid phosphatases and Pi transporters. N re-supply, but not P re-supply, represses genes assigned to phenylpropanoid and flavonoid metabolism.
We also compared the response to P and N deprivation on the individual gene basis. When comparing the set of 392 P-regulated genes (≥ 4-fold change between −P and +P conditions, FDR < 0.05; see previous section) with the set of 285 genes that display altered transcript levels between FN and N starvation (≥ 4-fold change, FDR < 0.05) in the data set from Scheible et al. (2004), an overlap of 59 N- and P-regulated genes results (Supplementary Fig. S2). Of the 392 genes that respond to P, 333 (85%) do not respond to N. Of these 333 P-specific genes, 30 and 46% show a > 75 and > 50% transcript reversion after 3 h Pi re-addition to P-deprived seedlings (Supplementary Fig. S2a). This contrasts with the 59 genes that respond to both N and P; of these, only 7 and 17% recover by > 75 and > 50% 3 h after re-supplying Pi. (Supplementary Fig. S2b). A similar trend was seen when a less stringent criteria (twofold change, FDR < 0.05) were used to define N- and P-responsive genes; in this case, 215 of 1058 P-responsive genes also responded to N deprivation, and 31% of the P-specific genes but only 16% of the N- and P-responsive genes recovered by > 50% after re-supplying Pi (Supplementary Fig. S2c,d). This analysis confirms that the P- and N-deprivation responses include many shared genes, and that the shared genes, on average, respond more weakly and slowly than the P-specific genes when P is re-supplied. This supports the idea that rapid reversion provides a useful additional criteria to identify genes that respond specifically to P (cf. Fig. 3b).
Response of specific functional categories, pathways and gene families to Pi
To refine the analysis to the single gene level, the ATH1 data were subsequently either manually extracted (e.g. for specific gene families, see subsequent discussion) or automatically analysed using the data visualization tool MapMan (Thimm et al. 2004; Usadel et al. 2005; http://gabirzpdde/projects/MapMan/). After RMA computation of expression values, the ratios in the biological replicates were averaged and converted to a log 2 scale and imported into MapMan as xls files (Supplementary Table S4). MapMan converts the values to a false colour scale and displays them on diagrams; transcripts that increase, decrease or change less than a given threshold are shown in blue, red and white, respectively. In the following MapMan plots, gene expression levels in FN medium or after Pi addition are expressed relative to the level in P-deprived seedlings, so an increase after P re-supply shows up as a blue colouration, and a decrease in +P compared to −P as a red colouration. A scale was selected where a 0.6- and 3-fold change on a log 2 scale is required to produce a visible colouration and colour saturation, respectively. The full set of expression data from the nine ATH1 arrays in this study is deposited as a Microsoft Excel spreadsheet (Supplementary Table S1).
Transcripts for many genes cannot be reliably measured by Affymetrix technology because of their low expression levels (Czechowski et al. 2004; Usadel et al. 2005) or because the probe sets are unspecific (ATH1 probe set extension _s_at), or the genes are not represented on ATH1 arrays. We used qRT-PCR to extend and confirm the information gathered from the ATH1 analysis. For this purpose cDNA samples were prepared from shoots and roots of Pi-replete and Pi-deprived seedlings. The separation into shoots and roots also gives additional information as compared to the ATH1 data which were obtained from whole seedlings.
Rapid, strong and specific coordinated regulation of a subset of genes required for Pi uptake
As expected (Smith et al. 2003), several genes involved in Pi transport were induced by P deprivation and were rapidly repressed after P addition (Supplementary Figs S3a & S4). Most of the strongly induced genes belonged to the Pht1 family. One gene (At1g73220) that, according to The Arabidopsis Information Resource (TAIR) annotation could encode a Pi or sugar transporter, was the seventh most strongly P starvation-induced gene in seedlings (Table 1). Other strongly responding genes included GPT2;1 and At3g47420, one of a pair of genes annotated as glycerol-3-phosphate permease (Table 1, Supplementary Fig. S4) (see further discussion).
Re-analysis of the nine members of the Arabidopsis Pht1 Pi transporter gene family (Mudge et al. 2002) and four additional Pi transporters (Pht2;1, Pht3;1-Pht3;3) by qRT-PCR extended the information obtained from ATH1 arrays (Fig. 5a). For example, it shows that Pht1;1, Pht1;2 and Pht1;4 are predominantly expressed in P-deprived roots, whereas Pht1;5 and Pht1;7 are much more abundant in P-deprived shoots. The high signal recorded by the unspecific ATH1 probe set 266184_s_at in P-deprived seedlings hence is due to the high and predominant expression of Pht1;4 in roots and the high and predominant expression of Pht1;7 in shoots. The qRT-PCR analysis also reveals that Pht1;7 and also Pht1;8 and Pht1;9, all three expressed at very low levels in Pi-replete conditions, are in fact the most strongly induced Pht genes in P-deprived plants (i.e. the threshold fluorescence cycle number drops by 10–17).
The response is quite specific for transporters related to Pi (Supplementary Fig. S4). Transcripts for Nrt2;1 (At1g08090) and Nrt1;1 (At1g12110) were ∼ 3-fold decreased in P-starved seedlings and three- to fourfold increased 3 h after Pi re-addition. There were also only small changes in the expression of genes encoding enzymes in the pathways of NO3- and ammonium assimilation (Fig. 6). Several sulphate transporters were weakly induced in P-starved seedlings but were unaffected by Pi re-addition (Supplementary Fig. S4). In the sulphate assimilation pathway, two genes encoding 5′-adenylylsulphate (APS) reductase (At4g04610, At4g21990) and one encoding APS kinase (At4g39940) were weakly induced in P-starved seedlings but were unaffected by Pi re-supply (Fig. 6). A group of genes putatively involved in divalent metal transport, AtZIP9 (At4g33020), AtZIP4 (At1g10970), AtIRT3 (At1g60960) and AtHMA2 (At4g30110) were repressed in P-starved seedlings. Changes in the expression of genes involved in responses to metallic elements like Al, Fe and Zn during P deprivation were reported previously in rice (Wasaki et al. 2003) and more recently also for Arabidopsis (Misson et al. 2005). However, none of these genes recovered rapidly after Pi re-addition (Supplementary Table S1).
The response to Pi addition differs strikingly from the response to NO3- addition to N-deprived plants (Scheible et al. 2004), which led to induction of genes for NO3- and sulphate uptake and assimilation, but had no marked effect on genes for Pi transport.
PHO1-like and other SPX domain-encoding genes
PHO1 and its 10 homologues have been implicated in the internal transfer of Pi (see Introduction). These proteins have a conserved topology with an SPX tripartite domain in the N-terminal hydrophilic portion and an EXS domain in the C-terminal hydrophobic portion (Wang et al. 2004). Proteins containing SPX and EXS domains have been implicated in Pi sensing in yeast. The Arabidopsis genome encodes at least nine further proteins, in addition to the PHO1 family, that contain an SPX domain. Phylogenetic analysis indicates that these can be subdivided into three clusters (Wang et al. 2004). Because of their possible role in Pi transport or signalling, their response was also analysed in more detail (Fig. 5b & Supplementary Fig. S3b).
PHO1 (At3g23430) responded only slightly to changes in P availability. Its homologue PHO1;H1 (At1g68740) was strongly induced in seedlings during P deprivation (Supplementary Fig. S3b), while the others had ATH1 expression values too low to assess their response. Expression analysis of the PHO1 homologues by qRT-PCR confirmed the induction of PHO1;H1 during P deprivation and provided additional insights, including the preferential expression of PHO1 in roots consistent with its function in xylem loading of Pi, or repression of PHO1;H9 in P-deprived shoots (Fig. 5b).
Three other SPX domain-encoding genes (At5g20150, At2g45130 and At2g26660), which lie in a phylogenetic cluster of four genes and have no known function, were strongly induced by P deprivation and repressed after Pi re-supply (Supplementary Fig. S3b). Two of these belong to the most strongly induced genes found using ATH1 arrays (Table 1). The fourth gene in this cluster (At5g15330) displayed an opposite but weak response to P deprivation. Two further SPX domain-encoding genes (At2g38920 and At1g02860) constitute a third branch of the phylogenetic tree proposed by Wang et al. (2004). At1g02860 displayed a slight (∼ 2-fold) repression in P starvation, whereas At2g38920 was below reliable ATH1 detection (Supplementary Fig. S3b, insert). These two genes are annotated as C3H4-type RING finger proteins at TAIR (http://www.arabidopsis.org), possibly providing another link between P signalling and the ubiquitin/26S proteasome pathway, in addition to the PHO2 E2 conjugase (Bari et al. 2006). Expression analysis of all SPX domain-encoding genes by qRT-PCR confirmed the results from the ATH1 arrays (e.g. the induction of At5g15330 in Pi deprivation) and provided novel insights, including repression of At2g38920 in P-deprived roots or P starvation-induced expression of At4g11810, a gene absent from the ATH1 array.
Phosphatases and RNAses
Low P induces phosphatases and ribonucleases, which release P from metabolites in the plant or organic sources outside the plant. The ATH1 array contains 47 genes that are annotated as ‘acid and other phosphatases’ in the MapMan ontology, including 27 purple acid phosphatase (PAP) genes. Eleven PAP genes and two putative phosphatase genes (At1g17710 and At1g73010) were reversibly induced by Pi starvation (Supplementary Fig. S3c). The latter represents segmental duplicated twins and were the second and third most strongly P starvation-induced genes (∼ 100-fold) in our experiments (Table 1). Two PAP genes (AtPAP3 and AtPAP4) were repressed in Pi starvation, but this was not reversed after Pi re-supply. The strong response of PAP genes resembles that seen in Misson et al. (2005), and the strong induction of AtACP5 (At3g17790) and the somewhat weaker induction of AtPAP12 (At2g27190) confirms previous studies (del Pozo et al. 1999; Li et al. 2002). Although Wu et al. (2003) reported PAP4 (At1g25230) to be induced during Pi deprivation, the transcript of this gene decreased in our studies.
Some PAP genes, including AtPAP11 (At2g18130), which was shown to be inducible by P starvation (Li et al. 2002), are absent from the ATH1 array. Investigation of the expression of all 29 known PAP genes (Li et al. 2002) by qRT-PCR qualitatively confirmed many of the ATH1 results (Fig. 5c), but also revealed that AtPAP19, which is absent from the ATH1 array, is a very strongly inducible PAP gene family member in shoots. Several other PAP genes that show moderate responses to Pi starvation on the ATH1 array are revealed as highly responsive by qRT-PCR (AtPAP5, AtPAP11, AtPAP14, AtPAP23 and AtPAP25).
Three of the 22 ATH1 probe sets representing ribonucleases (MapMan Bin 27.1.19) were induced in P starvation (Supplementary Fig. S3c). These are At1g14220, the well known RNS1 (At2g02990), and RNS2 (At2g39780), a potential ortholog of which was shown to be P starvation induced in Antirrhinum (Liang et al. 2002). Inspection of 12 ribonuclease genes by qRT-PCR confirmed the strong induction of RNS1 (Fig. 5c), but revealed no additional transcripts that change with P availability (not shown).
Coordinated regulation of a subset of genes involved in organic acid synthesis
P deprivation leads to synthesis of organic acids (see Introduction and Fig. 2f). This will allow P to be recycled from phosphorylated glycolytic intermediates. Organic acids are also secreted to solubilize Pi in the soil. Most enzymes in central carbon metabolism are encoded by small- to medium-sized families. The overview provided by MapMan (Fig. 6; see also Supplementary Fig. S5 for the changes at the level of the individual genes) showed that single members of the small gene families that encode phosphofructokinase (PFK) and Pyrophosphate:fructose-6-phosphate phosphotransferase (PFP), two members of the phosphoglycerate mutase family, two members of the large pyruvate kinase family and two members of the phosphoenolpyruvate carboxylase (PEPCase) family are induced by Pi deprivation. Two phosphoenolpyruvate carboxylase kinase (PPCK) genes (At1g08650 and At3g04530), which are required for the phosphorylation and activation of PEPCase (Hartwell et al. 1999; Fontaine et al. 2002), were strongly activated by P deprivation (Table 1). All these changes were partially reversed within 3 h after re-supply of Pi (Fig. 6b & Supplementary Fig. S5b). The induction of pyruvate kinase has been reported in P-deprived Arabidopsis (Wu et al. 2003) and rice (Wasaki et al. 2003), and a higher transcript level of PEPCase was shown in P-deprived lupin (Johnson, Vance & Allan 1996), but the induction of PPCK during Pi stress has not been previously described. Genes encoding other enzymes in glycolysis, and genes encoding enzymes of the oxidative pentose phosphate pathways and the tricarboxylic acid cycle were unaffected by Pi availability. In the mitochondrial electron transport chain, genes encoding non-phosphorylating NADH dehydrogenases, a δ subunit of the F1-ATPase (At5g12420) and an uncoupling protein were induced in low P, but these changes were not reversed after Pi re-supply (Fig. 6 & Supplementary Fig. S5).
The specific induction of genes for selected enzymes in P deprivation contrasts to the response to N, where depletion leads to widespread repression and NO3- re-addition to coordinated induction of a large number of genes that encode enzymes for almost every reaction required to convert carbohydrates into organic acids (Scheible et al. 2004). Organic acids are required during NO3- assimilation to act as counter anions, and to provide acceptors for amino groups (see Scheible et al. 2004 for further discussion).
It has been suggested that P deprivation stimulates carbohydrate synthesis, allowing P to be recycled from phosphorylated intermediates (Plaxton 1996; Abel et al. 2002). ADG2 (At5g19220), APL3 (At4g39210), GBS1 (At1g32900), GPT2 (At1g61800), encoding two large ADP-glucose pyrophosphorylase (AGPase) subunit, a granule-bound starch synthase and a plastid glucose-6-P transporter, respectively, were induced in P-starved seedlings and repressed by Pi re-supply (Fig. 6). This appears consistent with a switch towards starch synthesis in low P and starch breakdown in high P, as mirrored by the starch levels (Fig. 2f). On the other hand, two genes involved in starch breakdown, BAM5 (At4g15210) encoding β-amylase and GWD3 (At4g24450) encoding a glucan water dikinase, were also induced in P-starved seedlings (Fig. 6). One specific isoform of sucrose phosphate synthase (SPS4, At4g10120) was strongly induced, and two (SPS5a and SPS5b) were weakly induced, and one isoform of sucrose phosphate phosphatase (SPP1, At1g51420) was strongly induced in P-starved seedling. This transcriptional activation of sucrose synthesis in low P was reversed by Pi addition (see next discussion for more data). It may complement the well-characterized allosteric activation of SPS by low Pi (Stitt, Huber & Kerr 1987). One member of the sucrose synthase family (SuSy3, At3g43190) and three putative invertases were induced in low P and repressed after Pi re-supply. Two members of the fructokinase gene family (At3g54090 and At1g69200) were repressed by P depletion and induced after Pi re-supply. Unexpectedly, no hexokinases were repressed.
There are similarities but also differences to the response to N (Scheible et al. 2004). APL1, which was unaffected by P, was repressed during N deprivation and induced after NO3- addition (Scheible et al. 2004). While the two fructokinases that respond to P respond in a similar manner to N, several additional invertases and hexokinases are also induced after NO3- re-supply. This is consistent with a role for these genes in the mobilization of starch and sugars in response to signals from N metabolism, but not from Pi.
Amino acid and nucleotide metabolism
N depletion leads to a general repression of genes required for amino acid synthesis and induction of genes required for amino acid degradation, which is partially reversed within 3 h of NO3- re-addition (Scheible et al. 2004). P depletion led to more restricted changes. This included induction of GDH1 (At3g03910), GDH2 (At5g07440) and ASN1 (At3g47340), which resembles the response to low carbon (Price et al. 2004; Thimm et al. 2004). Genes assigned to the terminal stages of phenylalanine and tryptophan synthesis were induced in P deficiency (Fig. 6a) but this was not reversed after re-supplying Pi (Fig. 6b). This correlates with the increase of aromatic amino acids in P-deprived seedlings (see Fig. 2).
P deprivation led to coordinated changes in the expression of genes involved in nucleotide metabolism, including weak but widespread repression of genes for purine and pyrimidine synthesis, nucleotide salvage (e.g. adenine phosphoribosyltransferase At1g80050) and nucleotide metabolism (e.g. adenylate kinase At5g35170, uridylate kinases At3g18680, NDPK2 At5g63310), and the induction of genes involved in nucleotide breakdown [e.g. pyrophosphatases At3g53620, At4g29690 and At4g29700, or 3′(2′)5′-bisphosphate nucleotidase At5g64000] (Fig. 7a). These changes were reversed by P re-supply (Fig. 7b), with transcripts for the biosynthetic pathways reaching higher levels than in FN. The changes in expression of genes involved in nucleotide metabolism are more marked for P than N (cf. Scheible et al. 2004). The weak response of amino acid and the strong response of nucleotide metabolism to P reflects the importance of nucleotides as a major pool and source of P during Pi limitation.
Hewitt et al. (2005) reported that Pi limitation leads to increased (up to 10-fold) expression of several genes involved in pyrimidine synthesis and salvaging, including aspartate transcarbamoylase (ATCase), carbamoylphosphate synthase (CPSase), uridine monophosphate (UMP) synthase in Arabidopsis. Our data and those of Misson et al. (2005) do not support these results obtained with semi-quantitative RT-PCR. In addition, using qRT-PCR, we found that transcripts for CPSase (At1g29900 and At3g27740), uracil phosphoribosyl transferase (UPRTase) (At3g53900), UMP synthase (At3g54470) and UMP kinase (At3g60180) decreased by two- to fivefold (i.e. 1.0–2.5 qPCR cycles) in shoots and roots of Arabidopsis seedlings following P deprivation (Supplementary Fig. S5), confirming our ATH1 results.
Coordinate P regulation of genes required for nucleic acid and protein synthesis
P depletion and re-supply modified the expression of genes involved further downstream in P use, for example, in the synthesis of DNA, RNA and processes that depend on them (Fig. 8). P deprivation led to a weak but broad repression of many genes involved in cell division and the cell cycle, DNA synthesis, RNA synthesis (see later for the impact on transcription factors), RNA processing, amino acid activation and protein synthesis (Fig. 8a) and Pi re-supply led within 3 h to a weak but widespread reversal of most of these changes (Fig. 8b). This coordinated response would not be readily apparent from analyses that identify the most strongly induced genes, because the individual changes in expression are not large. The general decrease of the levels of many amino acids from 8 h onwards (Fig. 2) is consistent with a gradual activation of protein synthesis. This response has similarities and differences to the response to N. In both cases, nutrient deprivation leads to a repression and re-supply of the missing nutrient to induction of many genes involved in protein synthesis. However, whereas central amino acids fall in N-deprived plants (Scheible et al. 2004), they remain unaltered or rise in P-deprived plants (see previous discussion). In both cases, nutrient re-supply leads to a general decrease of amino acids after 8 h.
There were some exceptions to the general trend to induction of genes for nucleic acid and protein synthesis. One was the plastid ribosomal proteins, where the repression in low P was not reversed when Pi was re-supplied (see further discussion). Some individual genes showed specific and opposing changes, for example, an S6 ribosomal protein kinase (At3g08720) was induced in low P and repressed after Pi re-supply. The same gene shows a contrary response during N deprivation and NO3- re-addition (Scheible et al. 2004). In animals, S6 kinase is a downstream target of TOR (target of rapamycin) and is implicated in the regulation of translation in response to growth factors, nutrients, stress and cellular energy levels (see, e.g. Hansen et al. 2005; Inoki et al. 2005; Soliman et al. 2005).
Reprogramming of lipid metabolism
Härtel, Dörmann & Benning (2000) and Kelly et al. (2003) observed that P deprivation induces genes required for galactolipid synthesis, and proposed that Pi is conserved by replacing phospholipids by galactolipids during P starvation. This replacement occurs not only in chloroplasts, but also in the plasmalemma in the tonoplast (Andersson et al. 2005). In addition to galactolipids, glucosylceramides and sterolglycosides also increase in plasma membranes during Pi starvation (Andersson et al. 2005).
One of the striking features of the global response to P starvation in our experiments and those of Misson et al. (2005) is the strong induction of a set of genes required for or potentially involved in galacto- and sulpholipid synthesis. This included four mono- and digalactosyldiacylglycerol synthase genes (MGDs and DGDs), as well as SQD1, (At4g33030) and SQD2 (At5g01220) (Fig. 6a; Misson et al. 2005). Our results demonstrate that these changes are rapidly reversed after Pi addition (Fig. 6b). Three genes encoding UDP-Glc 4-epimerase (At4g23920) and UDP-galactose 4-epimerase (At1g10960 and At2g34850), which are potentially required to provide UDP-galactose for galactolipid synthesis, were reported to be weakly induced (two- to fourfold) in midterm and long-term P deprivation by Misson et al. (2005). In our study, At4g23920 and At4g10960 were only weakly induced (∼ 2-fold) in P-starved seedlings; the latter failed our robustness criteria, and At2g34850 was unaffected indicating that these might be secondary effects.
Substitution of galactolipids for phospholipids requires degradation of the latter to recycle fatty acids and glycerol-3-phosphate. As already noted by Misson et al. (2005), P starvation induces phospholipase D ζ2 (At3g05630), and gives an increased signal for an ATH1 probe set representing two phospholipases C (At3g03530/At3g03540) (Supplementary Fig. S3d). Our results demonstrate that these changes were almost totally reversed within 3 h after Pi re-supply. In-depth qRT-PCR expression analysis (Fig. 5d) revealed that both of the phospholipase C genes were strongly up-regulated. However, other phospholipase genes (e.g. At4g34930, At4g38690) were repressed by P starvation. Our study confirmed repression of At1g73600 encoding phosphoethanolamine N-methyltransferase, but did not confirm the induction of a homologous gene (At3g18000) (cf. Misson et al. 2005).
It is also possible that phospholipids are degraded via glycerophosphodiesterases (GPDEs). The response of these genes were not analysed in Misson et al. (2005). Our results reveal that four GPDE genes (At3g02040, At5g08030, At1g74210 and At5g41080) were reversibly induced by 3- to 30-fold in P-deprived seedlings (Supplementary Fig. S3d). As previously noted, the most strongly P-modulated transporter was a putative glycerol-3-P permease (GPT2; see Supplementary Fig. S4, Table 1). There was no change in expression of genes annotated as glycerol-3-P dehydrogenase (Fig. 6, Supplementary Table S1). These results indicate that low P leads to the rapid release and turnover of glycerol-3-P, possibly in connection with the replacement of phospholipids with other polar lipids. Other striking features were the reversible twofold induction of a gene involved in steroid synthesis (At3g03310) and the induction of several lipases including At1g08310, At3g44520 and At4g00500 under P limitation (Supplementary Fig. S3d). None of these genes were induced in low N (Scheible et al. 2004). P-deprived plants also showed altered expression of many genes involved in fatty acid synthesis and desaturation, but in this case, the changes did not revert quickly when Pi was re-supplied, indicating that they may be secondary.
T-DNA knockout mutants in two phospholipase C genes (At3g05630 and At3g03530) and the GPDE gene At5g08030 had lipid profiles indistinguishable from wild type in P-deprived and P-replete conditions (Gaude & Dörmann, personal communication; Nakamura et al. (2005). None of these genes has a segmental duplicated homologue, nor are other homologues expressed at comparable levels during P starvation. This may indicate considerable redundancy in the routes for phospholipid turnover or, alternatively, functions beyond membrane lipid reconstruction during P starvation.
P-dependent transcriptional changes related to photosynthesis and secondary metabolism probably are secondary
P depletion repressed numerous genes involved in the light reactions, chlorophyll synthesis, the Calvin cycle and photorespiration (Fig. 6a). There was a weak repression of phosphate/triose phosphate translocator (TPT) (Supplementary Fig. S4a) and the cytosolic fructose-1,6-bisphosphatase (FBPase) (Fig. 6a), which catalyse the export of triose P from the chloroplasts and the first dedicated step of photosynthetic sucrose synthesis. This contrasts with the induction of members of the SPS and SPP families, which are involved in the later and generic steps of sucrose synthesis (see previous discussion). The response of the genes involved in photosynthesis mirrors the response of the genes encoding plastid ribosomes (compare Fig. 6 with Fig. 8). None of these transcripts increased after 3 h P re-supply, suggesting that the transcriptional repression of photosynthesis may be a secondary response linked to lower demand for photosynthate and higher sugar levels during P limitation. A similar situation was found after depletion and re-addition of N (Scheible et al. 2004).
P depletion induced a large number of genes involved in terpene, phenylpropanoid and flavonoid metabolism, including many genes involved in anthocyanin synthesis (Fig. 6a). P depletion also induced genes involved in aromatic amino acid synthesis (see previous discussion; Fig. 6a). None of these changes reverted after P re-addition, indicating that they may be indirect. Although the response of secondary metabolism to P deprivation superficially resembled the response to N deprivation (Scheible et al. 2004), closer inspection reveals some differences. Firstly, different subsets of genes were regulated in the shikimic acid pathway, with low N leading to induction of the early stages of the pathway and the enzymes leading to phenylalanine (Scheible et al. 2004), whereas low P induced genes required for phenylalanine but also tryptophan synthesis (Fig. 6a). Secondly, a small subset of the changes of expression of genes involved in aromatic amino acid synthesis and secondary metabolism are reversed within 3 h when NO3- is re-supplied (Scheible et al. 2004), but not after adding Pi (Fig. 6b). Transcript levels for this small subset of genes were also reverted after adding NO3- to nitrate reductase-deficient Arabidopsis mutants (Wang et al. 2004); data were evaluated in the supplementary material of Scheible et al. (2004). These results indicate that while both P and N deprivation lead to massive indirect changes in secondary metabolism, NO3- also exerts a more direct action on the shikimate pathway and specific genes in phenylpropanoid and flavonoid metabolism.
Activities of enzymes in C and N metabolism
A robotized platform (Gibon et al. 2004) was used to investigate whether the changes of transcripts in P-deficient seedlings lead to changes in the maximal activities of 22 enzymes in C and N metabolism (Table 2). There was a significant 3.6-fold increase of PEPC activity and a two- to threefold increase of the activity of two key enzymes for sucrose synthase (cFBPase, SPS). There were smaller but also significant increases of glutamate dehydrogenase (GLDH), glutamine synthetase (GS) and shikimate dehydrogenase activity. Most other enzyme activities were unaltered or showed only slight but non-significant increases. PFP activity showed a weakly significant decrease in activity. This was unexpected, because PFP activity has been reported to increase during P deficiency in other species (Plaxton 1996). The response to P contrasts markedly to the response in N-deprived plants (Table 2), where there were small but significant (P < 0.05) decreases in the activity of numerous enzymes involved in sucrose and starch synthesis, glycolysis (but not PEPC), NADP-isocitrate dehydrogenase (NADP-ICDH), and NO3- and ammonium assimilation.
|Enzyme||Response to P deficiency||Response to N deficiency|
|Activity –P/FN||P-value||Activity –N/FN||P-value|
Figure 9 summarizes measurements investigating whether these changes are reversed in the first 24 h after addition of Pi. The lower time line shows the activities at different times in full medium; the top line shows the activities in P-deficient medium, and the middle line shows the activities at different times after adding Pi. The activities are depicted using a false colour scale, with blue indicating an increase, and red indicating a decrease of activity relative to that in FN at the start of the experiment. A longer time course was taken than for the changes of transcripts, because changes of enzyme activities are often slower than changes of transcripts (Gibon et al. 2004). The increase of cFBPase activity was reversed within 8 h, and those of SPS, PEPC and GDH activity were partially reversed by 8–24 h.
Figure 10 compares the changes of enzyme activities with the changes of the transcript levels of genes that encode them. This analysis is restricted to the comparison of P-replete and P-deficient seedlings, where the changes were largest. The right-hand column depicts the change in activity of each enzyme, using a false colour scale. The left-hand column shows the changes of the corresponding transcripts. Most of the enzymes are encoded by two or more genes. The change in transcript level for each individual gene is depicted using a colour scale, and the level of the transcript for each gene by the length of the bar (see legend of Fig. 10 for details). The increase of SPS and PEPC activity in P-deprived seedlings was accompanied by an increase of the major transcripts from the corresponding gene families. The increase of GDH activity was accompanied by a marked increase of the two major transcripts (GDH1 and GDH2). GHD3, which is a minor transcript in these conditions, decreased. This contrasts with low N, where GDH3 was induced (Scheible et al. 2004). The weakly significant increases of the activities of invertase, glucokinase, fructokinase and nitrate reductase were accompanied by an increase of several transcripts or at least the major transcript of the corresponding gene families. However, the increase of GS activity was not accompanied by an increase of transcripts of the GS gene family; the highly significant increase of cytosolic FBPase activity was accompanied by a small decrease of cFBP transcript, and AGPase activity decreased slightly even though transcripts for AGPB and two of the four genes for AGPL increased (see previous discussion).
These results show that the higher activities of SPS, PEPC and GDH in P-deficient seedlings are at least in part due to transcriptional regulation. However, additional levels of regulation are involved in other cases including the cytosolic FBPase and probably AGPase. Notably, cytosolic FBPase activity reversed most rapidly after P re-supply (see previous discussion), and marked discrepancies between changes of transcripts and activity were recently noted for AGPase during diurnal cycles (Gibon et al. 2004).
Transcriptional regulators (TRs)
The list of robust P-regulated genes (cf. Fig. 3) was cross-checked against the genes that are annotated as TR in the MapMan ontology (MapMan Bin 27.3). Of the ∼ 1500 potential TR genes represented on the ATH1 chip, 31 displayed a robust (P-value < 0.05) and > 3-fold change in transcript abundance between −P and +P conditions (Table 3). These genes might therefore be regarded as candidates to orchestrate the transcriptional changes observed in response to altered P. The number of P-responsive TR genes and the magnitude of their response to altered P are much smaller than those observed in response to N (cf. Scheible et al. 2004). There were only two TR genes that displayed > 10-fold changes in transcript abundance between +P and −P conditions (Table 3). In addition, when NO3- is added to N-starved seedlings, several TR genes respond with > 10-fold changes within 30 min (Scheible et al. 2004). Such large and fast or transient changes were not observed for TR genes after Pi re-addition. To further define the specificity of the P response, we analysed the expression patterns of these genes using AtGenExpress data for abiotic stresses, biotic stresses, light and hormone treatments (http://www.arabidopsis.org/info/expression/ATGenExpress.jsp) as well as our own data for C, S and N deprivation (Table 3). Four of the TRs (At1g24260, At1g23140, At1g71130 and At2g34210) were totally specific for P, and the latter three also responded quickly (within 3 h) to Pi re-addition.
|Arabidopsis Gene Identifier (AGI)||Gene||TR family||Segmental duplication||ATH1 arrays (fold change)||MAS5 call||qRT-PCR (ΔΔCT)||qRT-PCR (fold change)||Other stimuli|
|At1g66390a||PAP2||MYB||–||26.6||0.51||−||+||+||−4.5||2.2||23.5||0.3||OS –N –S|
|At1g52890a||NAM||At3g15500||9.10||0.77||+||+||+||−2.7||1.1||10.5||0.6||ST OS DR BIO ABA|
|At1g10585a||bHLH||–||7.68||0.53||+||+||+||−1.4||1.5||5.0||0.7||WND BIO ABA|
|At1g56650a||PAP1||MYB||–||5.69||0.93||−||+||+||−3.1||1||7.5||0.5||ST OS DR –N –S|
|At1g68670||MYB/GARP||At1g25550||4.23||0.31||+||+||+||−2.8||2.5||4.4||0.3||ST –C –N|
|At1g23140||C2 domain||At1g70810||20.02||0.14||+||+||+||NR||None found|
|At2g43000||NAC||–||4.69||0.82||+||+||+||NC||ST OS CO BIO UV|
|At3g15500||NAC3||NAM||At1g52890||4.2||0.93||+||+||+||NC||ST MJ –C|
|At3g25730||AP2||At1g13260||3.19||0.76||+||+||+||NC||ST CO WND|
|At2g31230||ERF/AP2||At1g06160||3.18||0.55||+||+||+||NC||ST MJ BIO|
|At1g72830||HAP2C||CCAAT||3.1||1.1||+||+||+||NC||SEN SEED NO3|
|At1g68520||Zn Finger||At1g25440||0.31||1.62||+||+||+||NC||ST OS CO –C|
|At1g76890||GT2||Trihelix||0.26||1.34||+||+||+||NC||ST OS –C|
|At4g36540||bHLH||At2g18300||0.19||1.28||+||−||+||NC||ST OS CO –C|
|At2g18300||bHLH||At4g36540||0.14||1.72||+||−||+||NC||ST OS CO –C|
|At2g33720b||ABI3/VP1||–||1.01||0.97||−||−||–||3.0||−4.2||6.1||3.4||–N NO3 –S|
|Pi deprivation marker genes|
|At2g38940e||AtPt2||–||37.0||0.13||+||+||+||−5.3||3.5||24||0.1||ST CO BIO|
Two C2 domains containing TR genes (At1g23140 and At4g15740) were found in the set of 31 genes, and a third C2 domain gene (At1g70810, the segmental duplication of At1g23140) was only narrowly excluded (cf. Supplementary Table S1). The C2 domain is thought to be involved in calcium-dependent phospholipid binding and is found in many cellular proteins involved in signal transduction or membrane trafficking, including most plant phospholipases (Qin & Wang 2002). This is interesting in the light of the major changes occurring in the content and metabolism of membrane phospholipids during P starvation (see further discussion).
Several TRs showed interesting shared or opposite responses to N and P. PAP1 (AT1G56650) and PAP2 (AT1G66390) were strongly induced by P and also by N (Scheible et al. 2004) and S deprivation (Bielecka, Morcuende, Stitt & Scheible, unpublished results). Genetic studies suggest that PAP1 and PAP2 are involved in the transcriptional activation of anthocyanin and flavonoid biosynthesis (Borevitz et al. 2000). Presumably, these two genes contribute to the regulatory network that induces many genes involved in phenylpropanoid and flavonoid biosynthesis during nutrient deprivation. Three TR genes (At3g25790, At1g13300 and At1g68670) belonging to the (MYB-like) GARP family (Riechmann 2002) were responsive to P and N deprivation, as well as NO3- re-addition (Scheible et al. 2004). However, in this case, the response to P was opposite to the one to NO3- and N, and the magnitude of the response was much more pronounced for NO3- (up to 18-fold induction after 30 min re-addition; Scheible et al. 2004) than for P deprivation (three- to fourfold repression). The conserved stimulus-dependent expression of At3g25790 and At1g13300, a pair of genes that arose by segmental chromosome duplication, suggests considerable functional overlap between these two genes. Another member of the GARP family (PHR1, At2g20400) was previously reported to be involved in P homeostasis by regulating the expression of several P starvation-induced genes including AtIPS1, AtRNS1 and AT4 (Rubio et al. 2001). More recent data suggest that PHR1 has an important role in the regulation of many more strongly P starvation-inducible genes including miR399 encoding genes (Bari et al. 2006). Interestingly, PHR1 transcript did not change in response to P nutrition (cf. Table 3).
A real-time qRT-PCR platform (Czechowski et al. 2004) that measures transcript levels of most known TR genes with high sensitivity and precision was used (1) to confirm the results from the ATH1 arrays, (2) to search for P-responsive TR genes that are expressed at such low levels that they cannot be measured reliably using array technology, and (3) to investigate TRs that are absent from the ATH1 array. Table 3 summarizes the 20 TR genes we found reproducibly P responsive by qRT-PCR profiling of three additional biological replicates for −P, +P and 3 h Pi re-addition conditions (qRT-PCR results for Pi starvation inducible marker genes are shown for comparison at the bottom of Table 3). The selection criteria included a ΔΔCT value > 3 or < −3 or a calculated fold change > 4.0 or < 0.25 and a t-test P-value < 0.05 for all three biological replicates (cf. Supplementary Table S5). Twelve of the 31 TR genes identified with ATH1 arrays were clearly confirmed by qRT-PCR. These include the strongly P-responsive PAP1, PAP2, the MYB/GARP gene At3g25790 and the P-specific At2g34210, At1g71130 and At1g24260. The qRT-PCR study highlighted At2g34210 as one of the most Pi deprivation-induced TR genes (ΔΔCT value < −10). At2g34210 encodes a predicted ∼ 111 kD KOW domain-containing protein with a presumed role in the regulation of transcriptional elongation and translation. Interestingly, this gene is also located adjacent to the highly Pi-responsive microRNA399d, e and f genes (Bari et al. 2006) in the Arabidopsis genome.
qRT-PCR revealed an additional eight P-responsive TR genes that are either expressed at levels that escape Affymetrix technology or are absent from the ATH1 array (Table 3). These included MYB114 which is a tandem duplication of PAP2, the strongly P starvation-induced MADS box gene AGL47, two bHLH genes (ORG2 and ORG3), which were shown to be targets of the DOF-type TR OBP3 (Kang et al. 2003), and two homeobox genes (At3g03260 and At3g18010). Interestingly, At3g03260, which was very strongly induced by P starvation (ΔΔCT < −9, Table 3) contains a sterol/phosphatidylcholine-binding START domain, again pointing to a link between P signalling and lipid metabolism. Circumstantial evidence that phospholipid metabolism may contribute to P signalling is provided by the observation (see previous discussion) that T-DNA knockout mutants for two phospholipase genes and one glycerophosphodiester phosphodiesterase gene which are highly expressed during P starvation did not yield altered lipid profiles (Gaude & Dörmann, personal communication).
Seventeen TR genes identified with ATH1 arrays did not meet our stringent qRT-PCR selection criteria. Inspection of the qRT-PCR results in Supplementary Table S5 nevertheless confirms the responsiveness to P availability for six of them (i.e. At2g40750, At2g41070, At3g05690, At3g06490, At3g15500 and At5g28770), while for the residual 11 genes, qRT-PCR failed to confirm the ATH1 results. We consider it possible that the P response in Arabidopsis varies between experiments, experimenters and growth conditions (see subsequent sections for more evidence). The two C2 domain genes (At1g23140 and At4g15740) revealed by ATH1 profiling were not represented on the qRT-PCR platform; independent qRT-PCR results for At1g23140, however, confirmed its induction in P starvation (ΔΔCT ∼ 5, Supplementary Table S5).
Additional P-responsive genes involved in signalling and post-translational regulation
Using the annotation provided by MapMan, we next examined which other potential regulatory genes respond to P availability. MapMan BINs 29.4–29.5 contain 2349 genes with predicted functions in post-translational protein modification (protein kinases and phosphatases) and protein degradation (e.g. proteases, F-box proteins, E3 ligases), of which 1734 are represented by probe sets on the ATH1 array, and MapMan BIN 30 contains 1277 gene entries with functions in signalling, including receptor kinases, MAP kinases or calcium, G-protein, phosphoinositide and light signalling components, of which 1086 are represented by probes sets on the ATH1 array.
Transcripts for two tyrosine-specific protein phosphatases (At4g03960 and At1g05000) and a MAP2K kinase (At5g67080) were strongly increased during P deprivation and decreased markedly after 3 h Pi re-addition (Table 4). At5g67080's segmental duplicated twin, At3g50310, showed a similar but attenuated response. None of the many other MAP kinase or receptor kinase transcripts were significantly affected, and only a few other protein kinase transcripts (At4g23290, At4g11460, At1g72540, At4g14580, At5g01820) were changed by more than threefold (Table 4).
|Affymetrix probe set||Arabidopsis Gene Identifier (AGI)||TAIR annotation||Segmental duplication||ATH1 −P/+P||P30/−P||P180/−P||Specificity|
|255360_at||At4g03960||Tyr specific protein phosphatase||–||13.0||0.87||0.21||OS|
|265214_at||At1g05000||Tyr specific protein phosphatase||At2g32960||8.50||0.63||0.39||None found|
|252212_at||At3g50310||MAPKKK20||At5g67080||5.95||0.48||0.24||ST MJ UVB|
|254931_at||At4g11460||Protein kinase||At4g23300||0.15||0.90||1.52||ABA NO3|
|259921_at||At1g72540||Protein kinase||4.23||0.61||0.52||–N SEN|
|245563_at||At4g14580||CBL-interacting protein kinase 4||At3g23000||3.90||0.76||0.60||CO|
|251060_at||At5g01820||CBL-interacting protein kinase 14||3.85||0.74||0.92||ABA ST|
|247531_at||At5g61550||E3 ubiquitin ligase, protein kinase||–||8.96||0.69||0.29||None found|
|254097_at||At4g25160||E3 ubiquitin ligase, protein kinase||At5g51270||6.35||0.67||0.21||None found|
|253271_s_at||At4g34210 At4g34470||SKP1 E3 ubiquitin ligases ASK11/ASK12||9.82||0.48||0.32||ST MJ|
|267265_at||At2g22980||Serin carboxypeptidase||–||0.18||0.92||1.11||–C EN|
|267262_at||At2g22990||Serin carboxypeptidase||–||0.09||0.89||1.27||–C EN|
|251026_at||At5g02200||Related to PhyA specific signal transduction component||–||7.58||0.97||0.20||EN|
|253296_at||At4g33770||Inositol 1,3,4-trisphosphate 5/6-kinase||–||5.89||0.57||0.22||None found|
|255677_at||At4g00500||Calmodulin-binding triacylglycerol lipase||–||5.47||0.62||0.27||None found|
|247393_at||At5g63130||Phox/Bem1p domain-containing protein||At3g48240||14.3||0.23||0.14||ABA ST CO|
|246071_at||At5g20150||SPX domain-containing protein||–||48.5||0.53||0.09||OS CO|
|266132_at||At2g45130||SPX domain-containing protein||–||77.3||0.49||0.02||None found|
|267611_at||At2g26660||SPX domain-containing protein||–||5.09||0.63||0.29||None found|
|258887_at||At3g05630||Phospholipase D ζ2||–||23.1||0.73||0.15||OS ABA SEN SEED|
|259221_s_at||At3g03530 At3g03540||Phospholipases C2||–||55.7||0.82||0.10||ABA OS GT|
At5g63130 was found to be strongly P responsive and encodes an octicosapeptide/Phox/Bem1p domain-containing protein of unknown function. This domain is present in many eukaryotic cytoplasmic signalling proteins, including many NO3--responsive NIN-like transcription factors (Scheible et al. 2004). In addition, a gene (At4g33770) encoding inositol 1,3,4-trisphosphate 5/6 kinase was found to be P starvation induced. Besides their involvement in phytate biosynthesis, inositol polyphosphate kinases were recently found to be required for normal Pi sensing and root hair elongation (Stevenson-Paulik et al. 2005).
Surprisingly, few of the 1300 genes involved in targeted protein degradation via the ubiquitin/26S proteasome pathway in Arabidopsis (Smalle & Vierstra 2004) responded to P. None of the annotated Arabidopsis ubiquitin-conjugating E2 enzyme genes was affected by more than twofold. About ∼ 1200 loci encode E3 ubiquitin-protein ligases, including nearly 700 F-Box and about 500 RING-finger class proteins, (Kosarev, Mayer & Hardtke 2002; Smalle & Vierstra 2004). Of the 363 F-Box genes that we traced on ATH1 arrays, two (At1g23390 and At3g59940) were just about threefold repressed in P-deprived seedlings and none showed > 2-fold changes within 3 h of Pi re-addition. Among the 407 identified RING finger genes on the ATH1 array, four (i.e. At5g27420, At1g76410, At5g19430 and At4g11370) were three- to fivefold induced during P starvation, and two (At5g22920 and At1g22500) were repressed. Analysis of the 53 and 21 probe sets for U-box genes and SKP1 E3 ubiquitin ligase genes on the array revealed three (At5g61550, At4g25160, At4g34210/At4g34470) that gave 6- to 10-fold higher signals in Pi starvation and rapid reversion after Pi re-addition (Table 4). P also markedly affected the expression of four genes that encode proteases unrelated to the ubiquitin/26S proteasome pathway, including a subtilisin-like protease and three serin carboxypeptidases (Table 4).
Comparison with expression data in the public domain
The transcriptional response during P depletion has been previously analysed with 6–9K expression arrays in Arabidopsis (Hammond et al. 2003; Wu et al. 2003) and rice (Wasaki et al. 2003), and recently with 23K ATH1 GeneChips (Misson et al. 2005). The response to Pi re-addition response has also been studied sporadically, with less than 10 genes (Müller et al. 2004). Triplicate 23K ATH1 data sets are also available for a comparison of wild type and of pho1 mutant rosette leaves grown in Pi-replete conditions (http://affymetrix.arabidopsis.info/, NASCarrays, experimenter: J. Hammond). As the pho1 mutant has low shoot Pi levels (see Introduction), this provides a genetic system to analyse the long-term response to a suboptimal P supply in the leaves.
There is good agreement between the response to 2 d of P deprivation in our experiments and that in pho1 leaves (Supplementary Fig. S7). Many of the genes that are up-regulated in pho1 mutant leaves are also strongly up-regulated in P-deprived seedlings, and many of these also display a marked reversion in their transcript levels after Pi re-addition, confirming that they are responding to Pi. It is interesting that the pho1 mutant shows a good match, even though there is a long-term decrease of P in the shoot. One possible explanation may be that a continual supply of Pi to the shoot of pho1, albeit at a low rate, maintains the leaves at an intermediate P status.
Of the 866 genes that Misson et al. (2005) reported to show > 2-fold increased or decreased transcript levels during short-, medium- or long-term P deprivation in liquid culture- and agar plate-grown Arabidopsis, just 332 (∼ 38%) were represented in our set of 1058 robust P-regulated genes (cf. Supplementary Table S1, Fig. 3a). The agreement is surprisingly low considering the similar plant age and growth conditions in our experiments and in their experiments. When only the set of 77 or 324 genes that Misson et al. (2005) identified as responding to short- and midterm P deprivation was considered, the agreement became better (60 and 58%, respectively) (Supplementary Fig. S8). Restriction of the comparison to the 69 genes that were most responsive (> 10-fold) to P deprivation in our studies revealed an overlap of ∼ 87%. Importantly, transcripts of many of these overlapping genes were characterized as quickly reverting after re-addition of Pi to P-deprived seedlings (Fig. 3b, Supplementary Fig. S8), confirming that the re-addition response is a valuable criterion.
Comparison of our results with those obtained previously with an 8K Arabidopsis Affymetrix array (Hammond et al. 2003) and a 6K EST microarray (Wu et al. 2003) revealed very little agreement (Supplementary Figs S9 and S10, and supplemental text). These discrepancies may arise when P deprivation is not achieved at the time of sampling or when changes in transcript abundance during P deprivation are documented without further analysis to distinguish between direct responses to Pi, and indirect effects due to slow growth, carbohydrate accumulation or other secondary effects. It is also possible that plant age, developmental stage and growth conditions have significant influence on the P-starvation response. While we analysed the response in 9-day-old seedlings grown with constant light in sterile liquid cultures, Hammond et al. (2003) and Wu et al. (2003) investigated the changes that occur in 4–5-week-old plants grown hydroponically in light/dark cycles.
Altogether, these comparisons indicate that there are differences in the nature of P-responsive genes between different experiments, experimenters and conditions, and demonstrate the necessity of independent comprehensive, and multileveled approaches to characterize the P-starvation response. Cross comparison reveals a core set of genes that can be considered as robust P responsive after filtering the available ATH1 data sets by several criteria, that is, responsiveness to P deprivation, to Pi re-addition and genetic modification (Supplementary Table S1).
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
P deprivation leads to widespread changes in gene expression. We used ATH1 expression data from independent experiments to identify > 1000 genes that showed > 2-fold altered expression in response to Pi depletion. While the nature of moderately (two- to fivefold) P-responsive genes considerably varies between experiments, experimenters and conditions (cf. Misson et al. 2005), a core set can be defined. Further, a (at least partial) reversion of gene expression after 3 h Pi re-supply serves as a useful criterion to identify direct responses to Pi availability.
The most strongly and reversibly P-responsive genes included various genes with ill defined/unknown functions (e.g. SPX genes), Pi transporters (and among these especially the ones that escape ATH1 technology), several acid phosphatases, genes involved in galactolipid synthesis and phospholipid breakdown, and a set of genes required for the synthesis of organic acids. The induction of the latter is well mirrored by the strong increase of PEPCase activity and metabolites such as PEP, citrate and malate in P-deficient seedlings. Smaller, coordinated changes in expression of genes involved in carbohydrate synthesis, and widespread and coordinated changes for many genes involved in nucleotide metabolism and nucleic acid and protein synthesis were observed in response to altered P. These changes too were confirmed at the enzyme and metabolite levels (e.g. increase of cFBPase and SPS activities, and sucrose and starch levels in P-deficient seedlings).
A set of genes involved in cellular signalling, including a small number of transcription regulators, a MAP2K kinase and several genes related to the targeted protein degradation pathway also shows a robust, strong and reversible response to P deprivation. The identification of such P-responsive regulatory genes represents an important step towards characterizing the sensing and signalling pathways that allow metabolism, cellular growth and development to adjust to changes in P supply. While this assumption is also fueled by the identification and functional importance of highly P-responsive microRNAs (Fujii et al. 2005; Bari et al. 2006), it also has to be noted that PHR1, a key regulator of Pi-starvation responses (Rubio et al. 2001; Bari et al. 2006), does not display a marked transcriptional response to Pi.
Many other genes that show marked changes of expression in P-deprived plants did not show a rapid reversion after P re-supply, indicating that they may reflect secondary responses to, for example, changes of carbohydrates, metabolism or slower growth. These include genes involved in photosynthesis and secondary metabolism, transport or assimilation of other nutrients and ions, and many putative signalling components.
The response to altered P availability shows some similarities to the response to N including a coordinated regulation of nucleotide, protein and nucleic acid synthesis, but also shows many differences. In particular, changes in lipid metabolism appear to be specific for P; changes in central carbon metabolism are more restricted and often opposite for P and N; changes in nucleotide metabolism are more pronounced, and those in amino acid synthesis are much less pronounced for P than N, and there is no evidence for specific changes in secondary metabolism in response to altered P. When regulatory genes are considered, some transcription regulators and also the SPX genes are highly specific for P, but others are shared with N and a range of abiotic stress responses.
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
The work was supported by the Max Plank Society and by the German Ministry for Education and Research (BMBF-GABI 0312277A). R. Morcuende is a recipient of a Ramón y Cajal grant from the Spanish Ministry of Science and Technology. Expert Affymetrix service was provided by the RZPD (German Resource Center for Genome Research, Berlin).
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
- 2002) Phosphate sensing in higher plants. Physiologia Plantarum 115, 1–8. , & (
- 2005) Phosphate-limited oat. The plasma membrane and the tonoplast as major targets for phospholipid-to-glycolipid replacement and stimulation of phospholipases in the plasma membrane. Journal of Biological Chemistry 280, 27578–27586. , , , & (
- 2006) pho2, a phosphate overaccumulator, is caused by a nonsense mutation in a microRNA399 target gene. Plant Physiology 141, 1000–1011. , , , , & (
- 2006) PHO2, microRNA399, and PHR1 define a phosphate-signaling pathway in plants. Plant Physiology 141, 988–999. , , & (
- 1994) The Arabidopsis ribonuclease gene RNS1 is tightly controlled in response to phosphate limitation. Plant Journal 6, 673–685. , , , , & (
- 2001) Root hairs confer a competitive advantage under low phosphate availability. Plant and Soil 236, 243–250. & (
- 1995) Controlling the false discovery rate – a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B 57, 289–300. & (
- 2000) Activation tagging identifies a conserved MYB regulator of phenylpropanoid biosynthesis. Plant Cell 12, 2383–2394. , , , & (
- 2000) Conditional identification of phosphate-starvation-response mutants in Arabidopsis thaliana. Planta 211, 13–22. , , & (
- 2004) Real-time RT-PCR profiling of over 1400 Arabidopsis transcription factors, unprecedented sensitivity reveals novel root- and shoot-specific genes. Plant Journal 38, 366–379. , , , & (
- 2005) Genome-wide identification and testing of superior reference genes for transcript normalisation in Arabidopsis. Plant Physiology 139, 5–17. , , , & (
- 1990) Subcellular compartmentation of uridine nucleotides and nucleoside-5′-diphosphate kinase in leaves. Plant Physiology 92, 637–641. , & (
- 1995) Characterization of a phosphate-accumulator mutant of Arabidopsis thaliana. Plant Physiology 107, 207–213. & (
- 2002) Galactolipids rule in seed plants. Trends in Plant Sciences 7, 112–118. & (
- 2002) Arabidopsis thaliana contains two phosphoenolpyruvate carboxylase kinase genes with different expression patterns. Plant, Cell & Environment 25, 115–122. , , & (
- 2004) The transcriptional control of plant responses to phosphate limitation. Journal of Experimental Botany 55, 285–293. , , , , & (
- 2005) A miRNA involved in phosphate-starvation response in Arabidopsis. Current Biology 15, 2038–2043. , , , & (
- 2004) A robot-based platform to measure multiple enzyme activities in Arabidopsis using a set of cycling assays, comparison of changes of enzyme activities and transcript levels during diurnal cycles and in prolonged darkness. Plant Cell 16, 3304–3325. , , , , , , , , & (
- 2002) Identification and characterization of the Arabidopsis PHO1 gene involved in phosphate loading to the xylem. Plant Cell 14, 889–902. , , , & (
- 2003) Changes in gene expression in Arabidopsis shoots during phosphate starvation and the potential for developing smart plants. Plant Physiology 132, 578–596. , , , , , , , , & (
- 2004) Genetic responses to phosphorus deficiency. Annals of Botany 94, 323–332. , & (
- 2005) Target of rapamycin-dependent activation of S6 kinase is a central step in the transduction of nutritional signals during egg development in a mosquito. Journal of Biological Chemistry 280, 20565–20572. , , & (
- 2000) DGD1-independent biosynthesis of extraplastidic galactolipids after phosphate deprivation in Arabidopsis. Proceedings of the National Academy of Sciences of the USA 97, 10649–10654. , & (
- 1999) Phosphoenolpyruvate carboxylase kinase is a novel protein kinase regulated at the level of expression. Plant Journal 20, 333–342. , , , , & (
- 2005) Effects of phosphate limitation on expression of genes involved in pyrimidine synthesis and salvaging in Arabidopsis. Plant Physiology and Biochemistry 43, 91–99. , , & (
- 2000) The role of inorganic phosphate in the development of freezing tolerance and the acclimation of photosynthetic carbon metabolism to low growth temperature is revealed by studies of pho mutants of Arabidopsis thaliana. Plant Journal 24, 383–396. , & (
- 2005) Signaling by target of rapamycin proteins in cell growth control. Microbiology and Molecular Biology Reviews 69, 79–100. , , & (
- 2003) Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Research 31, e15. , , , , & (
- 1966) A new micromethod for the colorimetric determination of inorganic phosphate. Clinica Chimica Acta 14, 361–366. & (
- 1996) Phosphorus deficiency in Lupinus albus. Altered lateral root development and enhanced expression of phosphoenol pyruvate carboxylase. Plant Physiology 112, 31–41. , & (
- 2003) Target genes for OBP3, a Dof transcription factor, include novel basic helix-loop-helix domain proteins inducible by salicylic acid. Plant Journal 35, 362–372. , , , & (
- 2003) Disruption of the two digalactosyldiacylglycerol synthase genes DGD1 and DGD2 in Arabidopsis reveals the existence of an additional enzyme of galactolipid synthesis. Plant Cell 15, 2694–2706. , & (
- 2004) Control of sulphate assimilation and glutathione synthesis, interaction with N and C metabolism. Journal of Experimental Botany 55, 1831–1842. & (
- 2002) Evaluation and classification of RING-finger domains encoded by the Arabidopsis genome. Genome Biology 3, RESEARCH0016. , & (
- 2005) Analysis of C and N metabolisms and of C/N interactions using quantitative genetics. Photosynthesis Research 83, 251–263. , & (
- 2002) Purple acid phosphatases of Arabidopsis thaliana. Comparative analysis and differential regulation by phosphate deprivation. Journal of Biological Chemistry 277, 27772–27781. , , , , , & (
- 2002) AhSL28, a senescence- and phosphate starvation-induced S-like RNase gene in Antirrhinum. Biochimica et Biophysica Acta 1579, 64–71. , , , & (
- 2004) Responses of primary and secondary metabolism to sugar accumulation revealed by microarray expression analysis of the Arabidopsis mutant, pho3. Journal of Experimental Botany 55, 1221–1230. & (
- 2002) Phosphate availability alters architecture and causes changes in hormone sensitivity in the Arabidopsis root system. Plant Physiology 129, 244–256. , , , , & (
- 1995) Root architecture and plant productivity. Plant Physiology 109, 7–13. (
- 1995) Mineral Nutrition of Higher Plants 2nd edn, Academic Press Limited, London, UK. (
- 2000) Influence of cytokinins on the expression of phosphate starvation responsive genes in Arabidopsis. Plant Journal 24, 559–567. , , , , , , & (
- 2001a) The diurnal changes of nitrogen metabolism in leaves of nitrate-replete tobacco are driven by an imbalance between nitrate reductase activity and the rate of nitrate uptake and ammonium and glutamine metabolism during the first part of the light period. Plant, Cell & Environment 24, 177–190. , , , , & (
- 2001b) Elevated carbon dioxide increases nitrate uptake and nitrate reductase activity when tobacco is growing on nitrate, but increases ammonium uptake and inhibits nitrate reductase activity when tobacco is growing on ammonium nitrate. Plant, Cell & Environment 24, 1119–1137. , , , , & (
- 2005) A genome-wide transcriptional analysis using Arabidopsis thaliana Affymetrix gene chips determined plant responses to phosphate deprivation. Proceedings of the National Academy of Sciences of the USA 102, 11934–11939. , , , et al. (
- 2005) The Arabidopsis SUMO E3 ligase SIZ1 controls phosphate deficiency responses. Proceedings of the National Academy of Sciences of the USA 102, 7760–7765. , , , et al. (
- 2002) Expression analysis suggests novel roles for members of the Pht1 family of phosphate transporters in Arabidopsis. Plant Journal 31, 341–353. , , & (
- 2004) Gene expression during recovery from phosphate starvation in roots and shoots of Arabidopsis thaliana. Physiologia Plantarum 122, 233–243. , , & (
- 2005) A novel phosphatidylcholine-hydrolyzing phospholipase C induced by phosphate starvation in Arabidopsis. Journal of Biological Chemistry 280, 7469–7476. , , , , & (
- 1996) Regulation of PHO4 nuclear localization by the PHO80-PHO85 cyclin-CDK complex. Science 271, 209–212. , , & (
- Temporal responses of transcripts, enzyme activities and metabolites after adding sucrose to carbon-deprived Arabidopsis seedlings. Plant Journal (in press). , , , et al.
- 1996) Organization and regulation of plant glycolysis. Annual Review of Plant Physiology and Plant Molecular Biology 47, 185–214. (
- 1991) A mutant of Arabidopsis deficient in xylem loading of phosphate. Plant Physiology 97, 1087–1093. , , & (
- 1999) Type 5 acid phosphatase gene from Arabidopsis thaliana is induced by phosphate starvation and by some other types of phosphate mobilising/oxidative stress conditions. Plant Journal 19, 579–589. , , , , , & (
- 2004) Global transcription profiling reveals multiple sugar signal transduction mechanisms in Arabidopsis. Plant Cell 16, 2128–2150. , , & (
- 2002) The Arabidopsis phospholipase D family. Characterization of a calcium-independent and phosphatidylcholine-selective PLD zeta 1 with distinct regulatory domains. Plant Physiology 128, 1057–1068. & (
- R Development Core Team (2005) R, a language and environment for statistical computing. Vienna, Austria. R Foundation for Statistical Computing. ISBN 3-900051-07-0. http://www.R-project.org.
- 1999) Phosphate acquisition. Annual Review of Plant Physiology and Plant Molecular Biology 50, 665–693. (
- 2004) Development and evaluation of an Arabidopsis whole genome Affymetrix probe array. Plant Journal 38, 545–561. , , & (
- 2002) Transcriptional regulation, a genomic overview (September 30, 2002). In The Arabidopsis Book (eds C.R.Somerville and E.M.Meyerowitz). American Society of Plant Biologists, Rockville, MD, USA. doi/10.1199/tab.0085, http://www.aspb.org/publications/Arabidopsis/ (
- 1994) The responses of plants to non-uniformous supplies of nutrients. New Phytologist 127, 635–674. (
- 2001) A conserved MYB transcription factor involved in phosphate starvation signaling both in vascular plants and in unicellular algae. Genes and Development 15, 2122–2133. , , , , , & (
- 2004) Sulfur assimilatory metabolism. The long and smelling road. Plant Physiology 136, 2443–2450. (
- 1997) Nitrate acts as a signal to induce organic acid metabolism and repress starch metabolism in tobacco. Plant Cell 9, 783–798. , , , , & (
- 2000) Reciprocal diurnal changes of phosphoenolpyruvate carboxylase expression and cytosolic pyruvate kinase, citrate synthase and NADP-isocitrate dehydrogenase expression regulate organic acid metabolism during nitrate assimilation in tobacco leaves. Plant, Cell & Environment 23, 1155–1167. , & (
- 2004) Genome-wide reprogramming of primary and secondary metabolism, protein synthesis, cellular growth processes, and the regulatory infrastructure of Arabidopsis in response to nitrogen. Plant Physiology 136, 2483–2499. , , , , , , , , & (
- 2004) The ubiquitin 26S proteasome proteolytic pathway. Annual Review of Plant Biology 55, 555–590. & (
- 2003) Phosphate transport in plants. Plant and Soil 248, 71–83. , , & (
- 2004) Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Statistical Applications in Genetics and Molecular Biology 3, 3. (
- 2005) The mammalian target of rapamycin signaling network and gene regulation. Current Opinions in Lipidology 16, 317–323. (
- 2005) Generation of phytate-free seeds in Arabidopsis through disruption of inositol polyphosphate kinases. Proceedings of the National Academy of Sciences of the USA 102, 12612–12617. , , , & (
- 2002) A plant for all seasons, acclimations in photosynthetic carbon metabolism during cold acclimation in Arabidopsis. Current Opinions in Plant Biology 5, 199–206. & (
- 1987) Control of photosynthetic sucrose synthesis. In The Biochemistry of Plants Vol. 10 (eds HatchM.D. & N.K.Boardman), pp. 327–409. Academic Press, New York, USA. , & (
- 1989) Determination of metabolite levels in specific cells and subcellular compartments of plant leaves. Methods in Enzymology 174, 518–552. , , & (
- 2004) MapMan, a user-driven tool to display genomics data sets onto diagrams of metabolic pathways and other biological processes. Plant Journal 37, 914–939. , , , , , , , , & (
- 2004) Arabidopsis pdr2 reveals a phosphate-sensitive checkpoint in root development. Plant Journal 37, 801–814. , , , & (
- 1998) An Arabidopsis mutant missing one acid phosphatase isoform. Planta 206, 544–550. & (
- 2005) Extension of the visualization tool MapMan to allow statistical analysis of arrays, display of corresponding genes, and comparison with known responses. Plant Physiology 138, 1195–1204. , , , et al. (
- 2004) Structure and expression profile of the Arabidopsis PHO1 gene family indicates a broad role in inorganic phosphate homeostasis. Plant Physiology 135, 400–411. , , & (
- 1993) Gene regulation by phosphate in enteric bacteria. Journal of Cellular Biochemistry 51, 47–54. (
- 2003) Transcriptomic analysis of metabolic changes by phosphorus stress in rice plant roots. Plant, Cell & Environment 26, 1515–1523. , , , et al. (
- 2001) Phosphate availability regulates root system architecture in Arabidopsis. Plant Physiology 126, 875–882. , , & (
- 2003) Phosphate starvation triggers distinct alterations of genome expression in Arabidopsis roots and leaves. Plant Physiology 132, 1260–1271. , , , , , & (
- 1999) Psr1, a nuclear localized protein that regulates phosphorus metabolism in Chlamydomonas. Proceedings of the National Academy of Sciences of the USA 96, 15336–15341. , , , & (
- 2001) pho3, a phosphorus-deficient mutant of Arabidopsis thaliana (L.) Heynh. Planta 212, 529–534. , & (
- 2006) Pyrimidine and purine biosynthesis and degradation in plants. Annual Review of Plant Biology 57, 805–836. , , , & (
- Top of page
- MATERIALS AND METHODS
- RESULTS AND DISCUSSION
- CONCLUDING REMARKS
- Supporting Information
Figure S1. Comparison of ATH1 results from the P-deprivation/Pi re-addition experiments. Figure S2. Comparison of Pi- and N-responsive genes, and distribution of the signal reversion frequency after Pi re-addition. Figure S3. ATH1 expression of various gene families in entire seedlings in P-replete conditions, during P deprivation and after 30 min and 3 h Pi re-addition. Figure S4. Expression of transporter genes. Figure S5. Expression of genes involved in glycolysis, the tricarboxylic acid cycle and mitochondrial electron transport; resolved to the single enzyme level. Figure S6. qRT-PCR expression of pyrimidine synthesis genes. Figure S7. Comparison of the transcriptional response in P-starved wild-type seedling shoots with the one in pho1 mutant leaves. Figure S8. Comparison of the P-starvation response reported by Misson et al. (2005) with the one in this study. Figure S9. Comparison of the Pi starvation response reported by Hammond et al.(2003) with the one in this study. Figure S10. Comparison of the Pi starvation response reported by Wu et al. (2003) with the one in this study. Table S1. ATH1 array data. Table S2. Metabolite levels. Table S3. Wilcoxon P-values and average changes for MapMan Bins/subBins. Table S4. MapMan Files Table S5. qRT-PCR results for 2200 transcriptional regulators. Table S6. qRT-PCR results and primers for P-responsive genes.
|Supplementary+Figure+S1.eps||267K||Supporting info item|
|Supplementary+Figure+S10.eps||2474K||Supporting info item|
|Supplementary+Figure+S2.eps||109K||Supporting info item|
|Supplementary+Figure+S3.eps||377K||Supporting info item|
|Supplementary+Figure+S4.eps||161K||Supporting info item|
|Supplementary+Figure+S5.eps||386K||Supporting info item|
|Supplementary+Figure+S6.eps||139K||Supporting info item|
|Supplementary+Figure+S7.eps||4885K||Supporting info item|
|Supplementary+Figure+S8.eps||232K||Supporting info item|
|Supplementary+Figure+S9.eps||163K||Supporting info item|
|Supplementary+Table+S1.xls||19761K||Supporting info item|
|Supplementary+Table+S2.xls||41K||Supporting info item|
|Supplementary+Table+S3.xls||223K||Supporting info item|
|Supplementary+Table+S4.xls||6420K||Supporting info item|
|Supplementary+Table+S5.xls||10096K||Supporting info item|
|Supplementary+Table+S6.xls||59K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.