| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online November 3, 2006; 10.1104/pp.106.090167 Plant Physiology 143:156-171 (2007) © 2007 American Society of Plant Biologists Genome-Wide Analysis of the Arabidopsis Leaf Transcriptome Reveals Interaction of Phosphate and Sugar Metabolism1,[W]Plant Biochemistry Laboratory, Department of Plant Biology, Royal Veterinary and Agricultural University, DK–1871 Frederiksberg C, Denmark (R.M., M.M., L.N., T.H.N.); Center for Biological Sequence Analysis, BioCentrum-DTU, Technical University of Denmark, Kemitorvet, DK–2800 Lyngby, Denmark (H.J.); and Department of Agricultural Sciences, Crop Science, Royal Veterinary and Agricultural University, 2630 Taastrup, Denmark (R.M.)
Global gene expression was analyzed in Arabidopsis (Arabidopsis thaliana) by microarrays comprising 21,500 genes. Leaf segments derived from phosphorus (P)-starved and P-replenished plants were incubated with or without sucrose (Suc) to obtain tissues with contrasting combinations of P and carbohydrate levels. Transcript profiling revealed the influence of the two factors individually and the interactions between P- and sugar-dependent gene regulation. A large number of gene transcripts changed more than 2-fold: In response to P starvation, 171 genes were induced and 16 repressed, whereas Suc incubation resulted in 337 induced and 307 repressed genes. A number of new candidate genes involved in P acquisition were discovered. In addition, several putative transcription factors and signaling proteins of P sensing were disclosed. Several genes previously identified to be sugar responsive were also regulated by P starvation and known P-responsive genes were sugar inducible. Nearly 150 genes were synergistically or antagonistically regulated by the two factors. These genes exhibit more prominent or contrasting regulation in response to Suc and P in combination than expected from the effect of the two factors individually. The genes exhibiting interactions form three main clusters with different response patterns and functionality of genes. One cluster (cluster 1) most likely represents a regulatory program to support increased growth and development when both P and carbohydrates are ample. Another cluster (cluster 3) represents genes induced to alleviate P starvation and these are further induced by carbohydrate accumulation. Thus, interactions between P and Suc reveal two different signaling programs and novel interactions in gene regulation in response to environmental factors. cis-Regulatory elements were analyzed for each factor and for interaction clusters. PHR1 binding sites were more frequent in promoters of P-regulated genes as compared to the entire Arabidopsis genome, and E2F and PHR1 binding sites were more frequent in interaction clusters 1 and 3, respectively.
Phosphorus (P) is an essential macronutrient and efficient acquisition of phosphate (Pi) is an important factor for plant growth. Consequently, plants have evolved a wide range of morphological and molecular adaptations to increase remobilization, uptake, and efficient use of Pi when availability is low. These adaptations require the plant to sense the level of Pi and change the expression of a number of genes accordingly.
In Arabidopsis (Arabidopsis thaliana), such P-responsive genes are exemplified by genes coding for RNases (Bariola et al., 1994
The use of microarray techniques in transcriptome analysis has opened new possibilities to elucidate the sensing, signaling, and regulatory pathways of the P starvation response. In a partial transcriptome analysis of responses to Pi starvation in Arabidopsis (Wu et al., 2003
Recently, comprehensive examination of global gene expression in response to P deficiency revealed coordinated induction of 612 genes and suppression of 254 genes, respectively (Misson et al., 2005
Evidence for close interactions between P- and sugar-sensing pathways has recently emerged. Aside from the crucial role of P in metabolic pathways (Paul and Stitt, 1993 These studies demonstrate that Pi influences sugar sensing of genes involved in C metabolism and also that sugars influence Pi-regulated genes with a function in P metabolism. However, the overall pattern and mechanisms behind these interactions still need to be outlined, and clearly this calls for closer examination. The focus of this study was to test the effect of P starvation and Suc accumulation in combination. Our hypothesis was that synergistic effects on gene expression by these two major nutrients have widespread importance for the regulation of P starvation-dependent genes, but also that further interaction patterns exist representing different regulatory programs, which will be revealed by analysis of global gene expression.
Previously, we demonstrated that resupply of Pi to starved plants is a useful approach to explore P starvation responses. Combination of this approach with sugar-feeding experiments makes it possible to control and change both Pi and sugar levels in tissues over short time periods, and this setup has proven valuable to study the interaction of P and sugar sensing (Müller et al., 2004
In this study, comprehensive analysis of gene expression in response to P starvation and Suc treatment was conducted using Arabidopsis Agilent microarrays with a total number of 21,500 genes represented. The aim was to explore cross-talk between transcriptional regulation in response to the two factors and to identify new candidate genes involved in P sensing.
Arabidopsis plants of similar size and development, but with different Pi status, were obtained by first growing plants on limited P supply (0.05 mM Pi in rockwool media) and then either maintaining these conditions or supplying the plants with high P concentration (4 mM Pi) for 1 week. Leaf segments with different Pi status were then incubated with or without Suc. Quantification of P in leaf segments verified that the total P content in resupplied plants was increased about 7-fold as compared to starved plants and the level of Pi was about 70-fold higher than in starved plants. As expected, leaves of P-starved plants had a higher level of sugars (3-fold) than P-supplied plants. During incubation, leaf segments remained floating, but with good contact to the solutions, and a considerable amount of exogenous Suc was taken up and metabolized to Fru and Glc (Table I ). These data confirm that the samples represent different combinations of Pi and sugar status.
General Features of the P Starvation and Sugar Treatment Expression Profile Two treatments at a time were compared on a set of slides revealing the relative changes in gene expression. With a significance level of P < 0.001, a total of 1,319 genes responded to P starvation (6.1% of the genes on the array), 5,479 genes responded to Suc (25.5%), and 149 genes exhibited alteration in transcript level, which revealed interaction between the two factors (0.7%). Full datasets for each of the factors P, Suc, and their interaction are available (Supplemental Tables S2, S3, and S4). There are both distinct and overlapping groups of genes regulated in response to P starvation and Suc (Fig. 1 ). Five hundred and five genes were significantly regulated in response to both treatments, but independently, whereas 72 genes, which also responded to both P and Suc, showed interaction between the two factors. This group of genes will respond to either of the factors alone, but the response pattern will also depend on the other factor. In principle, the combined effect of two factors without interaction could be assumed additive or multiplicative. When testing for interaction, we have assumed a multiplicative effect of the two factors as a basis for our evaluation because this is the more stringent test. A small group of 20 genes responded in a contrasting manner to both individual factors, depending on the other factor, and, therefore, on average there is no effect and the response is exclusively interaction. For example, if P depletion induces a gene at low Suc but represses the gene at high Suc, then there is on average no effect of P and the observed responses to P at high and low Suc are considered interaction.
Restricting the observations to those genes that changed expression by more than 2-fold, a total of 187 genes were regulated in response to P starvation (Fig. 1, sum of numbers in parentheses); of these, 171 genes were induced and 16 were repressed. These may represent useful targets in molecular breeding of crops for improved performance during P starvation. The list of P-responding genes (Supplemental Table S2) contains several new potential target genes compared to previous microarray studies (Hammond et al., 2003  ; Wu et al., 2003; Misson et al., 2005). Using the same 2-fold ratio as cutoff criteria, Suc incubation resulted in 337 up-regulated and 307 down-regulated genes.
Tight coupling exists between Pi and sugar metabolism and several studies have shown an interaction between sugar metabolism and Pi-regulated gene expression (Sadka et al., 1994 These data show that a large number of genes respond to both factors. However, for most of these genes (505), the two factors show no significant interaction, suggesting that these genes respond to simple metabolic connections. When analyzing the genes within this group that respond 2-fold or more, most of these (96%) are induced rather than repressed by sugar feeding. This is in contrast to the data for the total group of sugar-regulated genes, where 48% of the genes are repressed by sugar feeding. Thus, the group of 505 genes reacts more like the P starvation genes, where by far the majority of genes are induced (94%) rather than repressed by P starvation. We conclude that these genes, which would be characterized as sugar-sensing genes, are most likely responding to sugars indirectly through an effect on P metabolism. Importantly, more than one-half of the P-responding genes (742) were not influenced significantly by sugar incubation. This means that most P-responding genes were still independent from sugar treatment. Thus, the experimental procedure allows for observation of interactions other than simple P scavenging. The observed levels of Pi in the leaf tissues substantiate this because Suc incubation only had a limited effect on Pi level even in leaves with low Pi content. The experimental procedure allowed for observation of a group of 149 genes of which there is synergistic or antagonistic regulation in response to P and Suc. Of these genes, 37 showed more than a 2-fold change in expression. To our knowledge, most of these genes have not previously been described as both P and sugar responsive. The functional categories of the genes belonging to each of the three major Venn groups—P, Suc, and interaction—were obtained from The Arabidopsis Information Resource (TAIR; www.arabidopsis.org/tools/bulk/go/index) and the relative distribution between categories was compared to the whole Arabidopsis genome. In general, only small changes in distribution were observed at this level of organization (data not shown).
Transcriptional regulation revealed by microarrays was confirmed in a biologically independent experiment using reverse transcription (RT)-PCR (Figs. 2 and 3 ). Despite some quantitative differences in relative expression levels, RT-PCR data reflected the same regulatory patterns as the microarray data, showing clear induction or repression in response to Suc (Fig. 2). Similarly, induction of gene expression by P starvation was confirmed (Fig. 3), and here we also included a gene encoding a putative Tre-6-P synthase (At1g23870) that was found to be repressed in response to P starvation by both techniques. RT-PCR data confirm the expression data obtained by microarrays.
Genes Regulated in Response to Pi Starvation
In the following sections, we will mainly focus on P sensing and the interactions and less on sugar sensing because this has been covered by other studies using full Arabidopsis genome arrays and different experimental approaches (Koch, 1996
The genes regulated 2-fold or more in response to P starvation can be related to several functions, including P mobilization, signal transduction, transport, transcriptional regulation, and carbohydrate metabolism (Supplemental Table S3). Many of these 171 genes are directly linked to P mobilization and cleavage of Pi from P-containing compounds (e.g. phosphatases and ribonucleases), which agrees with the general perception of Pi-dependent gene regulation (Poirier et al., 1991
In Table II
, we have listed three groups of genes being regulated by P. One important group of genes encodes Pi transporters, which are classified into three families, Pht1, Pht2, and Pht3 (Rausch and Bucher, 2002
Other regulated genes are also related to P uptake and transport. The factor, PHF1, which facilitates correct membrane location of Pht1;1 (González et al., 2005 ), was induced (2.7-fold) in response to P deficiency, and this might support a faster processing of transporter proteins during increased rates of synthesis. Possibly the genes encoding the SPX domain-containing proteins (At5g20150, At2g26660) and the EXS family protein (At1g68740) are also related to P transport. The SPX domain has been identified in proteins, which appear to be involved in either transport or sensing of Pi (Hamburger et al., 2002; Wang et al., 2004). The SPX domain-containing proteins have some similarity to PHO1, a protein involved in xylem loading of Pi. The gene At5g20150 was 29-fold induced. Also, the EXS family protein contains a domain that relates to the PHO1 protein and has been classified as PHO1;H1 (Wang et al., 2004). The family comprises nine other homologs (PHO1;H2–10), but none of these responded to P starvation. Unfortunately, the PHO1 itself was not present on the array. Finally, two genes encoding an ATP-binding cassette transporter family protein and a putative germin-like protein are also connected to transport and storage.
Another important group of P starvation-induced genes codes for enzymes active in C metabolism (e.g. key enzymes of primary metabolism, including starch degradation, glycolysis, and Suc biosynthesis). Induction of such genes is likely to promote changes in C metabolism to improve P utilization and remobilization in the cell (Plaxton and Carswell, 1999
Of the 187 genes regulated more than 2-fold in response to P starvation, 11 transcription factors or putative transcription factors were identified (Table II). The expression level varied among these genes, but all levels were substantially above the limit of detection, especially at the combination low P and high Suc, where expression was maximal for all genes (Supplemental Table S5). In general, there is little information on transcription factors involved in regulation during P starvation. Until now, only two transcription factors have been conclusively shown to participate in P regulation in photosynthetic eukaryotes. These are the MYB family proteins, PSR1 from Chlamydomonas reinhardtii (Wykoff et al., 1999 ; Moseley et al., 2006) and PHR1 from Arabidopsis (Rubio et al., 2001). Two more related transcription factors appear to be associated with nutrient stress responses (Todd et al., 2004), and these factors belong to a family of 15 genes with homology to PHR1. In this study, none of the genes in this family, including PHR1, responded to P starvation. However, tissue localization is important because we know PHR1 is regulated in roots in response to P (L. Nilsson, unpublished data). Among the 11 transcription factors that we found to be regulated in response to P starvation, three encode MYB proteins. These are interesting candidates for P-regulating factors, although they do not belong to the PHR1 family. The most regulated (4.5-fold) of the transcription factor genes encodes an AP2 domain protein (At1g71130), which belongs to a group of factors with a highly conserved DNA-binding domain. Many of these factors belong to the ethylene-responsive element-binding protein group of transcription factors, several of which are known to be responsive to environmental stresses (Weigel, 1995). The AP2 domain-containing transcription factor was also observed in a recent microarray study (Misson et al., 2005). Finally, we find 11 P starvation-induced genes encoding proteins likely to be involved in signal transduction. These include a cornichon family protein (At4g12090), a putative calcium-dependent protein kinase (At1g08650), and a putative receptor kinase (At4g31240; Table II). We further analyzed the genes listed in Table II for coexpressed gene networks using the ATTED-II database (http://www.atted.bio.titech.ac.jp), which integrates the expression pattern from a large number of microarray studies. Several genes that were highly coregulated with the genes listed in Table II were also identified as P-regulated genes in this study. For the two most regulated factors, AP2 and CCAAT-binding transcription factor (At3g05690), we found that, respectively, six and eight of the 10 most correlated genes belonged to the P-regulated genes in our study. Thus, the lists of coexpression revealed by ATTED-II also extend to expression during P starvation, which suggests that these two factors are good candidates to be directly involved in regulating a part of the P response. Table III shows the 40 most up-regulated genes and their fold change in response to P starvation. This list includes a number of the above-mentioned genes with direct relation to Pi metabolism and six genes with unknown function.
The vast majority of the genes responding to Pi are up-regulated, whereas only 16 of 187 genes were down-regulated in response to P starvation (Supplemental Table S2). No common function is evident from the small selection of repressed genes. Apparently, there are only few dispensable reactions that are turned off to improve use of P in the leaf. One interesting response is the reduction of a light-harvesting complex II protein. This could represent a mechanism to reduce light harvesting to alleviate overreduction of photosystems as a consequence of Pi limitation. However, there are several light-harvesting complex isoforms and none of the others were responding to Pi treatment. Two other genes in the list are NMT1 and NMT3, which encode enzymes, phosphoethanolamine N-methyltransferases, involved in phospholipid biosynthesis; their regulation suggests specific reduction of the formation of phospholipids, thereby complementing the mechanisms for up-regulation of alternative lipids during P starvation.
In two previous array studies, Hammond et al. (2003)
Analyzing 1,000-bp upstream promoter regions using Athena Web tools (O'Connor et al., 2005
The frequency of motifs previously suggested to be mediating P responsiveness of genes was analyzed for all P-regulated genes showing more than 2-fold induction or repression by P starvation, respectively, and this was compared to the frequency in the whole Arabidopsis genome (Table IV
). For induced genes, the only both frequent and considerably enriched motif compared to the entire genome was the PHR1 binding site (Rubio et al., 2001
New potential cis-elements were searched using the promomer program (Toufighi et al., 2005 ; http://bbc.botany.utoronto.ca/ntools/cgi-bin/BAR_Promomer.cgi), which will identify enriched nucleotide sequences of 4 to 10 bp. One element, AGTTTT, appeared to be enriched in 16 P-repressed genes. Evaluating whether this is a true cis-element must await experimental analysis. Similar analysis of the induced genes revealed the motif GAATAT. Evaluating longer motifs and aligning all the sequences that included this core suggested that the motif could be further extended to ANGAATATNC. This motif includes the PHR1 recognition site, which validates that this unbiased approach can identify frequently occurring cis-elements. It further indicates that PHR1 might have a preference for an A in the second position of the binding site (now N) and for an additional A 2 bp upstream. Several other motifs were enriched, but these elements did not give similarly consistent results when different motif sizes were tested. Further cis-elements can potentially be revealed by more extensive approaches (e.g. as outlined by Geisler et al. [2006], who discovered several new Suc-responsive elements by an iterative method).
In plants, sugar accumulation induces the expression of genes involved in the synthesis of polysaccharides, storage proteins, pigments (e.g. anthocyanins), and genes associated with defense responses and respiration (Koch, 1996
The three genes most strongly induced in response to Suc encode a Glc-6-P/phosphate translocator precursor (GPT; At1g61800) induced 8.8-fold, a CHS up-regulated 6.6-fold (At5g13930), and a glycosyl hydrolase family 1 protein (At1g52400) induced 6.5-fold. Normally, GPT is not highly expressed in photosynthetic tissue, but it can be induced by stress conditions and, in the pho3 mutant, GPT is strongly induced (Lloyd and Zakhleniuk, 2004
Interaction of both factors is defined as a change in expression level in response to the two factors together, which is significantly higher or lower than expected from the changes in response to the two factors applied individually. In total, 149 genes displayed a highly significant interaction between the two factors (Fig. 1). Microarray data were verified by RT-PCR for 11 of the genes showing interaction, chosen to represent diverse functionality and strong regulation (Fig. 5 ). Most of these provided clear examples of expression patterns where induction by P starvation is much stronger in the presence of high sugar level. In addition, two of the genes (At3g02820 and At1g44900) show a different pattern, with higher expression when both P and sugar are ample.
To obtain an overview of the transcriptional pattern for these genes, clustering analysis was conducted. Log2 values of expression intensity ratios for each of the four comparisons were used as a basis for hierarchical cluster analysis (Eisen et al., 1998 ; http://rana.lbl.gov). The genes organize into three main clusters (Fig. 6
), where clusters 1, 2, and 3 comprise 66, 36, and 47 genes, respectively. These genes are also listed with their expression levels in Supplemental Table S4. Further categorization with respect to molecular function of the three clusters reveals large differences in the function of these three groups of genes (Fig. 7 ). The total group of 149 genes only showed minor deviation from the entire genome (data not shown), but this covers up that the interaction group includes different types of responses. When clustered according to their expression pattern, the genes also organize into different functions. Cluster 1 covers a relatively diverse group of genes, with clear overrepresentation of genes with structural molecule activity (9.5%), which were all coding for ribosomal subunits, and less frequently kinase, transporter, and hydrolase activities (7% compared to 22% for the entire genome). Cluster 1 is also enriched in DNA or RNA, protein, nucleic acid, or other binding (in total 42%), when compared to genes belonging to cluster 2 or 3. In contrast, genes belonging to cluster 2 and even more pronounced cluster 3 have a notably higher representation of kinase, transporter, hydrolase, and transferase activity. Together, these functional groups amount to 53% for cluster 2 and 70% for cluster 3, as compared to only 17% of the genes in cluster 1 (Fig. 7). To some degree, the clusters reflect which of the Venn groups (Fig. 1) the genes belong to as indicated by the color code in Figure 6.
To allow for overall comparison of expression pattern in each cluster, the level of expression was normalized for each gene, with the highest expression level set to 1 and the average expression pattern was plotted for genes that were regulated more than 2-fold (Fig. 8 ). For cluster 2, the two major subclusters (Fig. 6) were plotted separately because these showed a clearly different pattern. Genes in cluster 1 are induced by Suc feeding, but more so when the tissue has enough of both nutrients, P and sugar (Fig. 8A). Two examples of genes belonging to cluster 1 (Fig. 5, top images) clearly illustrate this pattern. Considering that the genes of cluster 1 are diverse in function and notably enriched in ribosomal subunits, we suggest that this cluster represents a molecular program to support increased growth and development when nutrients, in general, are ample. This agrees well with the notion that the function of many of the gene products of cluster 1 can be associated to cell growth, protein synthesis, regulation of the cell cycle, and DNA replication.
A search for overrepresented promoter motifs in cluster 1 revealed that, of 66 genes, 11 had promoter sequences that contained a short motif TCCCGC similar to the E2F binding-site motif (TTTCCCGC). The high frequency in cluster 1 contrasts with absence from clusters 2 and 3 and low frequency in the Arabidopsis genome. Table V shows enrichment of both the short motif core and the full E2F binding site. In a genome-wide search of E2F binding sites within the Arabidopsis genome, Ramirez-Parra et al. (2003) identified over 180 potential E2F target genes. Many of the targets were cell-cycle-related genes, but also other functional categories, such as transcription, stress, defense, and signaling, were found. In accordance, in our study all genes in cluster 1, which have the E2F binding site in their promoter, are cell cycle-related genes.
Genes in cluster 2 are less consistent in their regulation and fall into two subclusters that have a different expression pattern (Fig. 8, B and C). The first group has a complex pattern, being induced by P starvation at low sugar but repressed at high sugar. The other group is more related to P starvation-induced genes because these are preferentially expressed at low P in combination with high sugar. Genes in cluster 3 are induced by P starvation and also by Suc feeding and more so at low P (Fig. 8D). The expression patterns of eight genes belonging to this cluster were also confirmed by RT-PCR (Fig. 5). This pattern is compatible with genes primarily regulated in response to P starvation, a response that is then accentuated by Suc. As expected from this pattern, cluster 3 comprises genes that are likely to be directly involved in P mobilization either by cleavage of P from P-containing compounds or by metabolic changes resulting in improved P utilization and remobilization. Examples of such genes are purple acid phosphatases (At3g52820, At2g18130), calcineurin-like phosphoesterases (At1g52940, At3g46120), and phosphoesterase family proteins (At3g03530). Indeed, most of the genes characterized as hydrolases in this cluster were phosphatases. Examples of genes coding for enzymes of carbohydrate metabolism are Suc phosphate synthase (At4g10120), BMY1 (At4g15210), pyruvate kinase (At5g63680), and PFK (At4g32840). In this cluster, we also find transporter and storage-associated genes, such as a P transporter (At3g48850), genes coding for germins and germin-like transporters (At5g39110, At5g39160, At5g39130), a Suc transporter (At1g73220), and an ATP-binding cassette transporter (At3g59140). Induction of these genes can also be assumed to improve P metabolism of the tissue.
In cluster 3, genes encoding ethylene-responsive proteins and GA-regulated proteins were regulated in response to P starvation, and an auxin- and two GA-regulated proteins were induced in response to the interaction of both factors. In roots, regulation via these plant hormones, especially ethylene and auxin, is likely to be involved in alteration of developmental processes in response to limited P (Lynch and Brown, 1997 In agreement with our observation that cluster 3 represents genes induced to alleviate P starvation, we also find that the PHR1 binding site is specifically enriched in this cluster (Table V). Searching for other known or new motifs has not revealed any further motifs unique to the interaction clusters. Plants will experience severe stress when limiting P is combined with conditions that promote accumulation of carbohydrates, for example, during high irradiation or low temperature. We suggest that the cooperative regulation of genes observed in cluster 3 is likely to have evolved to meet the need for alleviation of P starvation during such conditions. Thus, organization of the genes showing interaction in three clusters reveals a close relationship between functionality of the gene products and the expression pattern. These data outline the contours of at least two regulatory programs. One, represented by cluster 1, serves to control genes with relation to growth and development during ample nutrient conditions. Another program, represented by cluster 3, has a function in coordination of adaptation to P starvation and C metabolism. A possible third program, represented by cluster 2, seems related to cluster 3 (subcluster 2-2), but, in general, cluster 2 is not as well defined. In this study, we have identified possible cis-regulatory elements, but it has not been possible to establish any unique elements for the genes showing interaction, and this may suggest that known elements, such as the PHR1 binding site, are important also for the observed interaction pattern. We suggest that an important future research topic should be to experimentally investigate whether these coordinated genes possess separate promoter elements that can be related to sugar and P responsiveness, respectively, or whether changing one element will affect the response to both factors.
Plant Material and Cultivation Conditions
Seeds of Arabidopsis (Arabidopsis thaliana) ecotype Columbia were germinated on soil and, after 3 weeks, transferred to 40-mL rockwool cubes (Rockwool). Plants were placed in a growth chamber at 20°C, 70% relative humidity, and 120 µmol m–2 s–1 photosynthetically active radiation over an 8-h photoperiod and supplied with nutrient solution with limiting Pi concentration (0.05 mM) as described by Müller et al. (2004)
Pi and total P were extracted and quantified as described by Müller et al. (2004)
The microarray Agilent Arabidopsis 2 oligo was used in this study. The array contained 22,000 probes (60mers) comprising 21,500 genes (Agilent; www.home.agilent.com), which represent approximately 80% of all Arabidopsis genes. The experiment was designed as a two-factor experiment with P = phosphate, Suc = sucrose, resulting in four treatments: (1) +P, –Suc; (2) +P, +Suc; (3) –P, –Suc; and (4) –P, +Suc. Two treatments at a time were compared by cohybridizing differentially labeled mRNA from two samples on a set of slides. The comparisons conducted were (1) (+P, –Suc) versus (–P, –Suc); (2) (+P, + Suc) versus (–P, +Suc); (3) (–P, –Suc) versus (–P, +Suc); and (4) (+P, –Suc) versus (+P, +Suc). The experiment included four biological replicates and four technical replicates. The technical replicates included dye swaps. Each sample represented leaf material from five to 10 plants.
RNA was isolated according to Müller et al. (2004) Cyanine 3- or cyanine 5-labeled cDNA targets were synthesized with the Agilent fluorescent direct labeling kit (Agilent) using 20 µg total RNA per reaction as input. The labeled target cDNA was purified using Qiaquick PCR purification kit (Qiagen). The cyanine 3- and cyanine 5-labeled targets were combined and hybridized to the slides for 17 h at 42°C using the Agilent hybridization kit according to the manufacturer's instructions.
Slides were scanned (scanner GMS418M; WG), and images were analyzed using ImaGene software (BioDiscovery).
Statistical analysis was performed with the freeware program R, using BioConductor packages (BioConductor, 2005). Normalization was performed by the use of the Qspline method (Workman et al., 2002
To discriminate genes with significant changes in their expression in response to P starvation or Suc or the interaction of the two factors, two-way ANOVA was conducted (the P < 0.001 cutoff was chosen after evaluation of the permuted data in a Vulcano-plot [Knudsen, 2002
To verify the microarray data with an independent method, RNA was isolated and DNase treated, and RT-PCR was conducted according to Müller et al. (2004)
The following materials are available in the online version of this article.
We thank Jonas Müller for helpful Ruby scripts for data conversion. Received September 21, 2006; accepted October 17, 2006; published November 3, 2006.
1 This work was supported by the Danish National Research Foundation, Center for Molecular Plant Physiology, and by the Danish Research Council for Technology and Production Sciences.  The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Tom Hamborg Nielsen (thni{at}kvl.dk).
[W] The online version of this article contains Web-only data. www.plantphysiol.org/cgi/doi/10.1104/pp.106.090167 * Corresponding author; e-mail thni{at}kvl.dk; fax 45–35283333.
Andersson MX, Stridh MH, Larsson KE, Liljenberg C, Sandelius AS (2003) Phosphate-deficient oat replaces a major portion of the plasmamembrane phospholipids with the galactolipid digalactosyldiacylglycerol. FEBS Lett 537: 128–132[CrossRef][ISI][Medline] Bariola PA, MacIntosh GC, Green PJ (1994) Regulation of S-like ribonuclease levels in Arabidopsis. Antisense inhibition of RNS1 or RNS2 elevates anthocyanin accumulation. Plant J 6: 673–685[CrossRef][ISI][Medline] Beutler HO (1984) Monosaccharides and derivatives. In HU Bergmeyer, ed, Methods of Enzymatic Analysis, Vol VI. Verlag Chemie, Weinheim, Germany, pp 331–337 Burleigh SH, Harrison MJ (1999) The down-regulation of Mt4-like genes by phosphate fertilization occurs systemically and involves phosphate translocation to the shoots. Plant Physiol 119: 241–248 Ciereszko I, Johnsson H, Hurry V, Kleczkowski LA (2001a) Phosphate status affects the gene expression, protein content and enzymatic activity of UDP-glucose pyrophosphorylase in wild-type and pho mutants of Arabidopsis. Planta 212: 598–605[CrossRef][ISI][Medline] Ciereszko I, Johnsson H, Kleczkowski LA (2001b) Sucrose and light regulation of a cold-inducible UDP-glucose pyrophosphorylase gene via a hexokinase-independent and abscisic acid-insensitive pathway in Arabidopsis. Biochem J 354: 67–72[CrossRef][ISI][Medline] Czechowski T, Stitt M, Altmann T, Udvardi MK, Scheible WR (2005) Genome-wide identification and testing of superior reference genes for transcript normalization in Arabidopsis. Plant Physiol 139: 5–17 Daram P, Brunner S, Rausch C, Steiner C, Amrhein N, Bucher M (1999) Pht2;1 encodes a low-affinity phosphate transporter from Arabidopsis. Plant Cell 11: 2153–2166 del Pozo JC, Allona I, Rubio V, Leyva A, de la Pena A, Aragoncillo C, Paz-Ares J (1999) A 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 J 19: 579–589[CrossRef][ISI][Medline] Eisen MB, Spellman PT, Brown PO, Botstein D (1998) Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 95: 14863–14868 Essigmann B, Güler S, Narang RA, Linke D, Benning C (1998) Phosphate availability affects the thylakoid lipid composition and the expression of SQD1, a gene required for sulfolipid biosynthesis in Arabidopsis thaliana. Proc Natl Acad Sci USA 95: 1950–1955 Franco-Zorrilla JM, González E, Bustos R, Linhares F, Leyva A, Paz-Arez J (2004) The transcriptional control of plant responses to phosphate limitation. J Exp Bot 55: 285–293 Geisler M, Kleczkowski LA, Karpinski S (2006) A universal algorithm for genome-wide in silico identification of biologically significant gene promoter putative cis-regulatory-elements; identification of new elements for reactive oxygen species and sucrose signaling in Arabidopsis. Plant J 45: 384–98[CrossRef][ISI][Medline] Gibon Y, Blaesing OE, Hannemann J, Carillo P, Höhne M, Hendriks JHM, Palacios N, Cross J, Selbig J, Stitt M (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 |
TOP
ABSTRACT
RESULTS AND DISCUSSION
-glucosidase (Malboobi and Lefebvre, 1997
2-fold change in expression. The colors included are used to designate groups in 
