Global expression profiling of sulfur-starved Arabidopsis by DNA macroarray reveals the role of O-acetyl-l-serine as a general regulator of gene expression in response to sulfur nutrition

Authors

  • Masami Yokota Hirai,

    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1-33, Inage-ku, Chiba 263-8522, Japan,
    2. CREST, JST (Japan Science and Technology Corporation), Japan, and
    Search for more papers by this author
  • Toru Fujiwara,

    1. Department of Applied Biological Chemistry, Graduate School of Agricultural and Life Sciences, University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
    Search for more papers by this author
  • Motoko Awazuhara,

    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1-33, Inage-ku, Chiba 263-8522, Japan,
    Search for more papers by this author
  • Tomoko Kimura,

    1. Department of Applied Biological Chemistry, Graduate School of Agricultural and Life Sciences, University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
    Search for more papers by this author
  • Masaaki Noji,

    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1-33, Inage-ku, Chiba 263-8522, Japan,
    Search for more papers by this author
  • Kazuki Saito

    Corresponding author
    1. Department of Molecular Biology and Biotechnology, Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho 1-33, Inage-ku, Chiba 263-8522, Japan,
    2. CREST, JST (Japan Science and Technology Corporation), Japan, and
      For correspondence (fax +81 43 290 2905; e-mail ksaito@p.chiba-u.ac.jp).
    Search for more papers by this author

For correspondence (fax +81 43 290 2905; e-mail ksaito@p.chiba-u.ac.jp).

Summary

To investigate the changes in profiles of mRNA accumulation in response to sulfur deficiency, approximately 13 000 non-redundant Arabidopsis thaliana ESTs corresponding to approximately 9000 genes were analyzed using DNA macroarray. Three-week-old Arabidopsis plants grown on an agarose-solidified control medium were transferred to a sulfate-free medium and grown for 48 h for the analyses of sulfur-related metabolites and global gene expression profiles. Concentrations of sulfate, O-acetyl-l-serine (OAS), a positive regulator of sulfur deficiency-responsive genes, cysteine and glutathione (GSH) were determined. Plants transferred to sulfate-free media had reduced concentrations of sulfate and GSH, and OAS concentrations increased. Macroarray analysis revealed a number of genes, including APR2 and Sultr1;2, whose mRNA accumulation was increased by sulfur deficiency. Profiling was also carried out with plants treated with OAS under sulfate-sufficient condition. Scatter plot analysis revealed a positive correlation between the changes of expression levels by sulfur deficiency and by OAS treatment among the clones tested, suggesting that mRNA accumulation of a number of genes under sulfur deficiency is mainly controlled by OAS concentrations in tissues. It was also revealed that the sets of genes regulated under sulfur deficiency in leaves and roots differ considerably.

Introduction

Sulfur is a macronutrient essential for plant growth. Sulfur is taken up by plants in its inorganic sulfate form, followed by reduction into sulfide and incorporation into cysteine. Cysteine is further converted to methionine, or incorporated into GSH and proteins.

Sulfur assimilation and sulfur-amino acid biosynthesis are controlled both by feedback regulation of enzyme activities and regulation of gene expression involved in the sulfur assimilation pathway (Hawkesford, 2000; Leustek and Saito, 1999; Saito, 2000). When plants are exposed to sulfur deficiency, sulfate uptake and assimilation activities are de-repressed. In Arabidopsis thaliana, the expression of the genes encoding isoforms of sulfate transporter (Sultr1;1, Sultr1;2, Sultr2;1, Sultr2;2, and Sultr4;1), adenosine 5′-phosphosulfate reductase (APR13) and serine acetyltransferase (SAT-p) is upregulated (Gutierrez-Marcos et al., 1996; Takahashi et al., 1997, 2000; Vidmar et al., 2000; Yoshimoto et al., 2002). In siliques and old rosette leaves of sulfur-starved Arabidopsis and in sulfur-starved cultured cotyledons of soybean, the concentration of OAS, a direct precursor of cysteine biosynthesis, increases (Awazuhara et al., 2000; Kim et al., 1997, 1999). Exogenous application of OAS enhanced sulfate uptake and ATP sulfurylase activity in sulfur-sufficient cell suspension culture of maize (Clarkson et al., 1999). Addition of OAS to barley grown with adequate sulfur supply led to increases in sulfate transporter mRNA (HVST1) accumulation and sulfate uptake rates (Smith et al., 1997). The mRNA level of Sultr1;1 increased by application of OAS to sulfur-sufficient Arabidopsis (Hatzfeld & Saito, 2000). Feeding of OAS to nitrogen-deficient Arabidopsis led to an increase in mRNA levels of three APR isoforms, sulfite reductase, two cysteine synthase (CS) isoforms, and a SAT isoform (Koprivova et al., 2000). These facts suggest that OAS may be a positive regulator of sulfur assimilatory genes in response to sulfur nutrition in plants as previously found in bacteria (Kredich, 1993).

Moreover, OAS is considered to be a regulator of expression of seed storage protein genes. The gene encoding the β subunit of β-conglycinin, one of the major seed storage proteins of soybean, is upregulated at the transcriptional level under sulfur-deficient conditions (Fujiwara et al., 1992; Gayler and Sykes, 1985; Hirai et al., 1995; Kim et al., 1999; Naito et al., 1994). OAS application to immature soybean cotyledons cultured in vitro changed the pattern of seed storage protein accumulation similar to that seen under sulfur deficiency (Kim et al., 1999). OAS contents in immature siliques of Arabidopsis increased as the sulfate/nitrate ratio in the culture media decreased, which was in parallel with the change in the transcriptional activity of the β subunit gene promoter in seeds of transgenic Arabidopsis (Kim et al., 1999). These facts suggest that OAS is a key compound involved in regulation of seed storage protein gene expression in response to sulfur and nitrogen nutrition.

Despite a number of physiological investigations, molecular mechanisms underlying regulation of gene expression in response to sulfur nutrition are largely unknown. In particular, no report has been available for global gene expression profiles under sulfur-nutrition stress. In this paper, we report DNA macroarray analysis using sulfur-starved and OAS-treated Arabidopsis to investigate the possibility that OAS is a general regulator for global gene expression under sulfur-nutrition stress. From our results, we propose that OAS is a key compound that regulates not only a particular set of genes but also global mRNA profiles of sulfur-starved Arabidopsis.

Results

Changes in sulfur-related metabolites and Sultr1;1 expression

To clarify the early changes in the transcriptome in response to sulfur deficiency, Arabidopsis was subjected to relatively short-term sulfur deficiency. Arabidopsis plants were grown on solidified control medium (1.5 mm sulfate) for 3 weeks, and then transferred to sulfate-free (no sulfate) or OAS-supplemented media (1.0 mm OAS/1.5 mm sulfate) and grown for 48 h. Under these conditions, the plants did not show chlorosis or any significant change in their morphological characteristics (data not shown). However, sulfate contents were reduced significantly in leaves and roots (Figure 1a). OAS treatment led to a 1.5-fold increase in sulfate contents in leaves but the change was not evident in roots (Figure 1a). Nitrate and phosphate contents did not change significantly by these treatments (data not shown). OAS contents increased statistically significantly under sulfur deficiency and by OAS supplement in both leaves and roots (Figure 1b). Cysteine contents increased significantly in the roots supplied with OAS (Figure 1c). GSH contents tended to decrease under sulfur deficiency, and increased significantly in roots by OAS (Figure 1d). The expression of Sultr1;1 is known to be upregulated in the roots under sulfur deficiency (Takahashi et al., 2000; Vidmar et al., 2000), and in the current study real-time PCR analysis revealed that the expression of Sultr1;1 was upregulated about seven- and ninefold in the roots shifted to sulfate-free and OAS-supplemented medium, respectively (data not shown).

Figure 1.

Contents of sulfur metabolism-related compounds in leaves and roots used for macroarray analysis.

Sulfate, O-acetyl-l-serine (OAS), cysteine and glutathione (GSH) contents were measured in leaves and roots of control (C), sulfur-starved (–S) and OAS-treated (+OAS) plants. Average and SD of four to five replicates are presented. Values indicated by different letters are significantly different from each other at P = 0.05 according to two-sided t-test.

Mathematical verification of global gene expression profile

Total RNA was extracted from the leaves and the roots of the above-mentioned plant samples and global gene expression profile analysis by DNA macroarray was conducted. 33P-labeled single-stranded cDNA was generated from the RNA and used as a target. Hybridization signals were quantified using the program package Array Vision (Imaging Research, St. Catharine's Ont., Canada). After subtraction of local backgrounds, about 10% of raw signal intensities were equal to zero and eliminated for further analysis as described in Experimental procedures. The relative signal intensity (normalized signal intensity) was calculated as described in Experimental procedures. A typical scatter plot of relative signal intensities of sulfur-starved samples versus those of the control sample is shown in Figure 2. The distribution of the spots spread along the diagonal line, and there was no distortion in the shape of distribution. A linear regression fits to this plot (Figure 2), suggesting no ‘range bias’ in our data (Finkelstein et al., 2002). Table 1 shows the slopes and correlation coefficients of scatter plots for all combinations of control and treated samples. The average correlation coefficient of scatter plots was 0.83, suggesting that the quality of our data was sufficiently high for further analysis. In fact, known sulfur deficiency-responsive genes such as APR2 were identified to be upregulated by our analysis (see below). Sultr1;1 is not available on JCAA (The Japanese Consortium for Arabidopsis thaliana DNA Array) macroarray used in this study, because of a lack of expressed sequence tag (EST) likely due to relatively low expression level of Sultr1;1 under sulfur-sufficient condition.

Figure 2.

Scatter plot of relative signal intensity corresponding to each gene under sulfur-starved versus control conditions.

Relative signal intensity of each spot in control leaf sample (C) was plotted against that in sulfur-starved leaf sample (–S). The data for the spots on the membranes C and D were combined for this plot. Black line and red broken line represent the linear regression line (y = 0.85x + 2.15, R2 = 0.84) and the diagonal line (y = x), respectively.

Table 1.  Linear regression of normalized signal intensities between control and treated samples
SampleMembrane1st hybridization2nd hybridization3rd hybridization
Versus –SVersus +OASVersus –SVersus +OASVersus –SVersus +OAS
  1. Relative signal intensity of each spot in control sample was plotted against that in treated samples (–S, sulfur-starved sample; +OAS, O-acetyl-l-serine-treated sample) as described in Figure 2. The slope of linear regression line and the correlation coefficient (in parenthesis) are shown. A macroarray was composed of four membranes, A, B, C, and D (see Experimental procedures). Data obtained from the membranes A and B, and from C and D were gathered for these calculations.

LeafA and B1.05 (0.91)1.07 (0.88)0.83 (0.84)0.77 (0.89)
C and D0.85 (0.84)0.81 (0.83)0.85 (0.87)0.92 (0.90)0.88 (0.86)0.90 (0.86)
RootA and B0.82 (0.76)0.68 (0.68)
C and D0.85 (0.72)0.91 (0.77)

Comparison of global expression profiles in response to sulfur deficiency and OAS

In order to investigate whether OAS is a key metabolite for regulation of transcript accumulation under sulfur deficiency in a genomic scale, we compared transcript profiles in response to sulfur deficiency and OAS treatments. ESTs with the sufficient repetitive data for statistical analysis both in control and treated conditions were selected, and degrees of induction/repression by the treatments, that is, ratios of the average relative signal intensities of treated sample to those of control sample were calculated. The degrees of induction/repression under sulfur deficiency were plotted against those by OAS treatment (Figure 3). There was a positive correlation between the degree of induction/repression by sulfur deficiency and that by OAS treatment (R2 = 0.30 for leaf, 0.42 for root). This suggests that a large number of genes induced or repressed by sulfur deficiency were also induced or repressed by OAS.

Figure 3.

Scatter plot of changes in expression levels under sulfur deficiency versus by OAS treatment.

The horizontal axis shows ratio of the average relative signal intensities of sulfur-starved sample to those of control sample. The vertical axis shows ratio of the average relative signal intensities of OAS-treated sample to those of control sample. Red and green dots represent the ESTs significantly induced and repressed under sulfur deficiency, respectively. The blue dots indicate the rest of ESTs. Lines represent the linear regression line for all ESTs shown in red, green, and blue.

(a) Transcripts from leaves.

(b) Transcripts from roots.

Global gene expression profile under sulfur deficiency in leaves and roots

We then compared the responses to sulfur deficiency in leaves and roots. Ratios of the average relative intensities of sulfur-starved plants to those of control plants in leaves were plotted against those in roots (Figure 4). Apparently, the dots were distributed around the crossing point of x- and y-axes, and there was no statistically significant correlation between the degrees of induction/repression in leaves and roots. This indicated that the set of genes induced/repressed in leaves was different from that in roots. Similarly, the set of genes induced/repressed in leaves by OAS was also different from that in roots (data not shown).

Figure 4.

Scatter plot of changes in expression levels under sulfur deficiency in leaves versus in roots.

The horizontal axis shows ratio of the average relative signal intensities of sulfur-starved leaf sample to those of control leaf sample. The vertical axis shows ratio of the average relative signal intensities of sulfur-starved root sample to those of control root sample.

Genes significantly increased or repressed identified by statistical analysis

By using the statistical program Significance Analysis of Microarrays (SAM, Tusher et al., 2001; see Experimental procedures), we identified the particular genes significantly induced/repressed by sulfur deficiency from the global gene expression profile data. In this analysis, the false detection ratio (FDR) was set at equal to or less than 10%.

As shown in Table 2, 216 and 282 ESTs out of approximately 13 000 were identified as those significantly upregulated under sulfur deficiency in leaves and roots, respectively. These ESTs are shown as red dots in Figure 3(a,b). There was a positive correlation between the degree of induction by sulfur deficiency and that by OAS (red dots, R2 = 0.25 for leaves, and 0.62 for roots). These results suggest that a similar set of genes is induced significantly both by sulfur starvation and by OAS addition. OAS content in leaves did not increase so much by OAS supplement as by sulfur deficiency, although the change was statistically significant (Figure 1). This might result in a lower correlation coefficient in leaves than that in roots. The ESTs repressed significantly under sulfur deficiency in leaves (276 ESTs) and roots (690 ESTs) are shown as green dots in Figure 3(a,b), respectively. These ESTs tended to be downregulated also by OAS, although there was no significant correlation between the degrees of repression by sulfur deficiency and by OAS (R2 = 0.0005 for leaves, and 0.0027 for roots).

Table 2.  The number of the expressed sequence tags (ESTs) identified induced or repressed significantly by sulfur deficiency and O-acetyl-l-serine (OAS) treatment
 InducedRepressed
LeafRootLeafRoot
  1. The numbers of ESTs induced or repressed significantly by sulfur deficiency (–S) and OAS treatment (+OAS), identified by statistical analysis SAM (FDR ≦ 10%), were shown. *Indicates that these ESTs were selected only from the membranes A and B (see Experimental procedures).

–S216282*276690*
+OAS10419*548850*

ESTs identified as those induced or repressed significantly under sulfur deficiency in leaves or in roots were plotted again in the way similar to Figure 4. There was no significant correlation between the degrees of induction/repression in leaves and in roots (data not shown).

These results indicate the major role of OAS for inducible effect on gene expression in leaves and roots, being more notable in roots, although the sets of genes induced differ between those from leaves and roots.

Putative functions of the genes regulated under sulfur deficiency

Significance Analysis of Microarrays was also carried out with FDR less than 2%, and a total 1429 ESTs were identified as those regulated by sulfur deficiency or OAS supplement. Their annotations were investigated using the Database for Arabidopsis Research and Tools (DART, http://biochem.agr.nagoya-u.ac.jp/atgenome/index.html, in Japanese language), a web site for the analyses of JCAA macroarray (Table S1). Among them, 305 ESTs were induced or repressed in more than two experimental conditions out of four (sulfur deficiency and OAS treatments in leaves and roots). About 20% (60 ESTs) were the clones for unknown/putative/hypothetical proteins and 18% (56 ESTs) were with no annotation. Fourteen percent (43 ESTs corresponding to 36 genes) were for ribosomal protein, most of which were downregulated under these conditions. The rest (146 ESTs) are summarized in Table 3.

Table 3.  ESTs significantly regulated by sulfur deficiency and/or OAS in more than two experimental conditions
  1. Values indicate the degrees of induction/repression calculated as ratios of the average relative intensities of treated samples (–S, sulfur-starved plants; +OAS, O-acetyl-l-serine-treated plants) to those of control samples. Values in red and green represent statistically significant induction and repression, respectively (SAM, FDR < 2%).

inline image

The EST corresponding to 5′-adenylylphosphosulfate reductase (APR2), which is known to be upregulated under sulfur deficiency (Gutierrez-Marcos et al., 1996), was upregulated significantly under sulfur deficiency in leaves and roots (Table 3, see also Table 4).

Table 4.  ESTs corresponding to the genes encoding sulfate transporters and the enzymes for sulfur metabolism
  1. Values indicate the degrees of induction/repression calculated as ratios of the average relative intensities of treated sample (–S, sulfur-starved plants; +OAS, O-acetyl-l-serine-treated plants) to those of control samples. –, Indicates that the degree could not be calculated because the signals of control and/or treated samples were too weak to be detected. Values in red and green represent statistically significant induction and repression, respectively (SAM, FDR < 2%). The degrees of induction greater than 1.2 and those of repression smaller than 0.8 are indicated in orange and pale green, respectively.

inline image

Interestingly, genes involved in jasmonic acid (JA) biosynthesis were upregulated under sulfur deficiency (Table 3). AtLOX2 and OPR1 are the genes encoding two JA biosynthetic enzymes, lipoxygenase and 12-oxophytodienoate reductase, respectively. Several JA- or methyl JA (MeJA)-responsive genes were also included in the list (Table 3). ESTs corresponding to a formate dehydrogenase gene and SPE2 for arginine decarboxylase were upregulated in leaves. These ESTs are identical to those reported to be upregulated by MeJA (Sasaki et al., 2001). Vegetative storage protein gene is also known to be upregulated by MeJA (Berger et al., 1995). ATRR2 was downregulated by MeJA (Schenk et al., 2000). These results indicate that the response in gene expression in leaves to sulfur deficiency and OAS additions is close to JA biosynthesis and responses to JA.

Several ESTs were annotated as nucleic acid-binding proteins and transcriptional factors. DBGET search (GenomeNet, http://www.genome.ad.jp) against GenBank database revealed that AT3g14230 is an ethylene-responsive element-binding protein (EREBP)-like protein. AT2g21660 is highly homologous to Ccr2, which is upregulated by cold and drought stresses (Carpenter et al., 1994).

Other ESTs were classified into several groups: genes involved in resistance to pathogens, cell wall synthesis, response to abiotic stresses, carbon metabolism and respiration, photosynthesis, protein degradation, cell maintenance, and others.

The ESTs corresponding to the genes encoding sulfate transporters and enzymes for sulfur assimilation are listed in Table 4. Sultr genes, which are known to be upregulated under sulfur deficiency (Takahashi et al., 1997, 2000; Vidmar et al., 2000; Yoshimoto et al., 2002), were induced under sulfur deficiency in our macroarray experiment. APR genes, which are also known to be upregulated by sulfur deficiency (Gutierrez-Marcos et al., 1996), were upregulated by sulfur deficiency both in leaves and roots (Table 4). APR2 expression was repressed by OAS treatment only in leaves (Table 4). Koprivova et al. (2000) reported that OAS treatment led to a slight increase in APR2 mRNA levels only in roots of nitrogen-starved Arabidopsis, in which APR2 was repressed both in leaves and roots compared to nitrogen-sufficient plants. Our result was consistent with that of Koprivova et al. (2000) in the point that OAS application had different effects on APR2 expression in leaves and roots.

Discussion

Plants treated with sulfur starvation or OAS supplementation

In this study, we focused on early adaptive responses to relatively short-term sulfur deficiency. We intended to clarify the role of OAS in the regulatory mechanism of global gene expression profile in the context of sulfur deficiency stress. Three-week-old Arabidopsis plants were treated with sulfur deficiency and OAS supplementation for 48 h. It is known that under these growth conditions the expression of the genes encoding sulfate transporter isoforms were upregulated (Takahashi et al., 1997, 2000; Yoshimoto et al., 2002). The concentrations of sulfate and GSH decreased within 48 h after transfer to sulfur-deficient medium (Figure 1), indicating that during 48 h the internal nutrient status of sulfur had changed and plants initiated to change gene expression for adaptation. However, the concentrations of cysteine did not change significantly (Figure 1) and plant did not show any visible symptom in our case. Considering that cysteine content drastically decreased under longer sulfur-deficient condition (Nikiforova et al., 2003), 48-h treatment under sulfur deficiency seemed to be sufficiently short to elucidate the early response to sulfur deficiency.

Interestingly, cysteine and GSH contents increased in roots supplied with OAS where OAS content dramatically increased, which was consistent with our experience (Sogawa Y. and Fujiwara T., unpublished results). This result was also consistent with the reports that the availability of OAS is rate-limiting for cysteine biosynthesis (Harms et al., 2000; Saito et al., 1994).

Validity of data

In micro- and macroarray analyses, reliability of signals is a key issue. First, labeled target should be carefully prepared. We checked the quality of the target by calculating the ratio of the amount of longer cDNA to that of all cDNA synthesized. When the ratio is less than 90%, cDNA should not be used as a target (Sasaki et al., personal communication). Our target cDNA was synthesized using oligo(dT)12−18 primers and sometimes did not seem to be elongated to reach the 5′ end of template mRNA. This may be the basis for different results for several ESTs corresponding to the same gene in some of our experiments. For example, ESTs APD02b05 and APD19f09 cover the 3′ region and the 5′ region of APR2 gene, respectively. The former was upregulated significantly by sulfur deficiency in leaves and roots, whereas the latter did not change significantly, although similar tendency to the former was observed (Table 4). Alternative possibility is discussed below.

Second, the elimination of false-positives is the most important matter of array analyses. Although many useful normalization strategies have been developed to correct for the systematic biases, no single normalization method has become a standard (Finkelstein et al., 2002). SAM, developed by Tusher et al. (2001), assigns a score to each gene on the basis of change in gene expression relative to the standard deviation of repeated measurements and identifies genes with statistically significant changes in expression. A number of papers were published in which SAM was successfully used for data analyses (Blanchard et al., 2001; Detweiler et al., 2001; Furlong et al., 2001).

We also analyzed the data with Student's t-test using normal (i.e. not logarithmic) value data (local background subtracted, or not subtracted), and with SAM using the normal value data, to identify the clones induced/repressed significantly. The global induction and repression patterns obtained by three different analyses led to the same conclusions that: (i) there was a positive correlation between the degrees of induction/repression by sulfur deficiency and by OAS application both in leaves and roots; and (ii) there was no correlation between the degrees of induction/repression by sulfur deficiency in leaves and roots (see below).

Possibility of cross-hybridization

We observed that the degrees of induction/repression calculated from the signal intensities for several ESTs corresponding to a single gene were different to each other. This might result from cross-hybridization between the isoforms of gene families. As mentioned above, in case of APR2, APD02b05 was upregulated significantly by sulfur deficiency in leaves and roots, whereas APD19f09 did not change significantly (Table 4). The former clone includes the 3′ untranslated region which has low homology between the isoforms, whereas the latter covers only the coding region conserved in three APR genes to cause obscure result.

Global expression profile in response to sulfur deficiency

The main purpose of this study is to profile the global gene expression in response to relatively short-term sulfur deficiency, and to describe the trends of gene expression in response to sulfur deficiency and OAS. For this purpose, we plotted the degree of induction/repression by sulfur deficiency against that by OAS treatment. The positive correlation between them suggests that OAS mediates the expression of not only sulfur-assimilatory genes but also other genes in response to sulfur deficiency. However, this does not necessarily mean that OAS is the only compound important for regulation. Cysteine and GSH are known to be negative regulators of gene expression (Bolch et al., 1999; Lappartient et al., 1999; Vauclare et al., 2002; Vidmar et al., 1999, 2000). In our experiments, cysteine and GSH contents in roots treated by OAS increased (Figure 1). As shown in Table 3, some ESTs were upregulated by both sulfur deficiency and OAS treatment, whereas others were upregulated under sulfur deficiency and downregulated by OAS treatment. In the latter case, increased GSH might overcome the positive effect of OAS on gene expression. It is interesting that there is no EST in Table 3, which was repressed significantly under sulfur deficiency and induced significantly by OAS. Such an expression pattern could not be explained simply by OAS and GSH contents. We assume the several regulatory pathways of gene expression under sulfur deficiency. Some genes may be regulated via OAS, some may be via cysteine and/or GSH, and others may be regulated by the combinations of OAS, cysteine and GSH or by unknown factor(s). For example, the gene encoding the β subunit of β-conglycinin is shown to be regulated by the ratio of OAS and methionine levels (Hirai et al., 2002).

In OAS-supplemented leaves, the increase in OAS content was little compared to that in sulfur-starved leaves, although the change was statistically significant (Figure 1). However, the global transcript profile changed similarly to the case of sulfur-starved leaves. This observation depicts a possibility that OAS content in particular organelles might increase sharply. To better understand the regulatory mechanism, concentrations of metabolites should be determined in specific organelles or cell types where the regulatory mechanism works.

Putative functions of the genes regulated by sulfur deficiency and/or OAS

For identification of the genes by comparison of two or more experimental conditions, we used a statistical analysis SAM (see Experimental procedures) and identified a list of genes whose expression was regulated (Tables S1 and 3).

More than half of identified genes were downregulated under short-term (2 days) sulfur deficiency (Table 2). It is in contrast to the report by Nikiforova et al. (2003) where more genes were upregulated under relatively long-term deficiency. They also observed that the ratio of the number of the upregulated genes to that of the downregulated genes increased with increasing duration of sulfur deficiency (6–10 days). Drastic decrease in the environmental sulfate concentration resulted from medium exchange might cause the general and temporal downregulation of gene expression, and then specific genes might be upregulated afterward.

In our experiments, the genes involved in JA biosynthesis and those induced by JA or MeJA treatment were upregulated (Table 3). It is quite interesting that under different experimental conditions, Nikiforova et al. (2003) also found that JA biosynthetic pathway was induced. Recently, He et al. (2002) reported that exogenous application of JA caused premature senescence in the wild-type Arabidopsis. JA levels increased in senescing Arabidopsis leaves and genes encoding the enzymes that catalyze most of the reactions of the JA biosynthetic pathway were differentially activated during leaf senescence, suggesting that JA may play a role in the senescence program (He et al., 2002). Although our plant samples subjected to sulfur deficiency did not show any changes in appearance, it is reasonable to consider that senescing program mediated by JA had just started in these plants for redistribution of internal sulfur. In addition, OAS induced the same response. Interestingly, it is reported recently that sulfur deficiency-responsive APR1 gene is upregulated by MeJA and its precursor 12-oxo-phytodienoic acid (OPDA) (Sasaki-Sekimoto et al., 2003).

Almost all genes classified to protein degradation category are components of ubiquitin-dependent proteolytic pathway. Proteins destined for degradation are tagged by conjugation to ubiquitin, whose process is carried out by an enzyme complex consisting of an ubiquitin-activating enzyme (E1), an ubiquitin-conjugating enzyme (E2), and an ubiquitin-ligating enzyme (E3). E3 and other proteins assemble to form SCF (Skp1-Cullin-F-box protein) complex (Spremulli, 2000). Recently, SCFCOI1 ubiquitin-ligase complexes are shown to be required for the response to JA in Arabidopsis (Xie et al., 1998; Xu et al., 2002). SCF complex also plays an essential role in auxin response (for review, see Gray and Estelle, 2000). In our study, nitrilase 1 and auxin response factor 1 were also regulated (in others category). Nitrilases (NIT1, NIT2, and NIT3) are involved in auxin biosynthesis, and NIT3 promoter activity is strongly induced by sulfur starvation and OAS, suggesting that NIT3 plays a role in the regulation of root morphology in sulfur-starving Arabidopsis (Kutz et al. 2002). Nikiforova et al. (2003) also suggested the involvement of auxin in the sulfur starvation response. SCF complex-mediated JA and/or auxin responses might occur under sulfur deficiency. In relation to this, genes involved in defense response to pathogens, which is known to be mediated partly by JA, were identified among the clones regulated by sulfur deficiency and OAS.

AT1g13930, a putative hydroxyproline-rich glycoprotein (HRGP) gene was upregulated in leaves and downregulated in roots (Table 3, cell wall synthesis category). In Chlamydomonas reinhardtii, two cell wall-localized HRGPs, which are very poor in sulfur-containing amino acid, accumulated under sulfur deficiency as the substitutes for sulfur-rich cell wall components (Takahashi et al., 2001). Database search against GenBank and MAtDB (MIPS Arabidopsis thaliana database; http://mips.gsf.de/proj/thal/db/) revealed at least four other putative HRGP genes of Arabidopsis. Peptide for AT1g13930 has second least sulfur-containing amino acids (1.9%) among the five, implying the similar adaptive strategy in leaves to C. reinhardtii.

When C. reinhardtii was starved of sulfur, there was a dramatic reduction in the level of photosynthesis due to loss of functional photosystem II reaction centers (Wykoff et al., 1998). It may be the case with Arabidopsis. In our study, however, sulfur-sufficient plants which were grown in plastic plate did not seem to have sufficient photosynthetic activity, judging from the fact that they could not grow to maturity without exogenous sucrose supply (data not shown). It may be the reason why the genes for photosynthesis were downregulated mainly in roots.

Genes involved in sugar metabolism were also regulated. In leaves, At5g17310 (UDP-glucose pyrophosphorylase) was downregulated, and two glycolysis/gluconeogenesis-related genes AT3g04120 and AT2g21330 (glyceraldehyde-3-phosphate dehydrogenase C subunit and putative fructose bisphosphate aldolase, respectively) were upregulated. It may suggest that sucrose synthesis for carbon storage was repressed and glycolysis occurred to generate energy. However, Arabidopsis has a number of isoforms for each step of sugar metabolism, and each isoform may express in specific organs, tissues and organelles under specific regulation. In fact, the isoforms were regulated differentially by sulfur deficiency (data not shown), suggesting more complicated and fine regulation.

In our study, AT1g75280 (NADP oxidoreductase, putative) was upregulated in leaves and roots. This is one of the highest over-expressers in the study of Nikiforova et al. (2003) and they gave a good speculation on its function. Several other genes (AT1g19570, AT2g34420, AT5g54270, AT4g16190, and AT5g34850) were shown to be regulated significantly both in Nikiforova et al. (2003) and our (Table 3) experiments.

Compared to Nikiforova et al. (2003), the degrees of induction/repression were generally lower in our study (Table 3). There are several possible reasons for this fact. First, as mentioned above, concentrations of metabolites in the sulfur assimilation pathway were not so severely influenced in our plants compared to the case of Nikiforova et al. (2003) possibly resulting in mild change in gene expression. Second, lower fold change was due to our analytical method in which the data were converted to logarithm for normalization. We normalized the raw signal intensities after conversion to logarithmic values according to Equation (1) (see Experimental procedures). As MED was greater than 10, relative intensity(1) became smaller than relative intensity(2) calculated without conversion to logarithmic value according to:

image(2)

Then, the degree of induction/repression, calculated as the ratio of the average relative intensities, became smaller when using Equation (1). For example, the degrees of induction of AT3g45140 (AtLOX2) and At1g76680 (OPR1) in sulfur-starved leaves calculated by Equation (1) were 1.77 and 1.52, respectively (Table 3). On the other hand, when calculated by the Equation (2), the degrees were 2.95 and 2.24, respectively.

Experimental procedures

Plant growth conditions

Arabidopsis thaliana ecotype Columbia was grown vertically on a solidified control medium (1.5 mm sulfate) at 22°C under fluorescent light (16-h light/8-h dark) for 3 weeks. Control medium contained 1% w/v sucrose, vitamins for Gamborg B5 medium (Gamborg et al., 1968), and MS medium salts (Murashige and Skoog, 1962) modified by replacing sulfate salts except for magnesium sulfate with equivalent chloride salts. Medium was buffered with 2-(N-morpholino)ethanesulfonic acid (MES) at 0.5 g l−1 (pH 5.8) and was solidified by 0.8% w/v electrophoretic-grade agarose LO3 (TaKaRa, Kyoto, Japan). In each plastic plate (15 cm in diameter) with 100 ml medium, 15 plants were grown to avoid the shortage of nutrients. Three-week-old-plants were transferred to control, sulfate-free medium (no sulfate, prepared by replacing magnesium sulfate with magnesium chloride), or OAS-supplemented medium (prepared by addition of 1.0 mm OAS to control medium), and grown for 48 h. Plant samples were dissected into rosette leaves and roots on ice, and immediately frozen with liquid nitrogen and stored at −80°C until macroarray experiments and measurement of compounds.

Measurement of sulfur metabolism-related compounds

Rosette leaves and roots were homogenized in 5 and 10 volumes of 0.01 m HCl (fresh weight basis) with mixer mill MM 300 (Qiagen, Valencia, CA, USA), respectively. After centrifugation at 18 000 g for 10 min, supernatants were ultrafiltrated through ULTRAFREE®-MC 5000 NMWL Filter Unit (Millipore, Bedford, MA, USA) to remove proteins. Sulfate, nitrate, and phosphate contents were measured by capillary electrophoresis system Agilent HP3D CE (Hewlett-Packard, Palo Alto, CA, USA) with fused silica capillary (75 µm × 48.5 cm) using Inorganic Anion Buffer for HPCE (Hewlett-Packard) at 25°C. Voltage was set at −30 kV and signals at 350 nm were detected by photodiode array (reference 245 nm). OAS content was measured by HPLC system (Hitachi, Tokyo, Japan) according to Kim et al. (1997). OAS was detected by fluorescence spectrophotometry after post-column reaction with O-phthalaldehyde. Contents of thiols were measured by HPLC system according to Anderson (1985) and Fahey and Newton (1987). Forty microliters of extract was reduced by 3 µl of 20 mm DTT and 100 µl of 100 mm 2-(cyclohexylamino)ethanesulfonic acid (CHES) (pH 9.3) for 30 min at 37°C. Then, the reduced extract was reacted with 10 µl of 30 mm monobromobimane for 40 min at 37°C. The labeling reaction was terminated by the addition of 50 µl of acetic acid, and then the resulting solution was subjected to HPLC analysis. HPLC was carried out as previously described (Saito et al., 1994).

RNA extraction and 33P-labeled target preparation

Total RNA was extracted from rosette leaves and roots using RNeasy Plant Mini Kit (Qiagen) according to manufacturer's instruction. For preparation of 33P-labeled target cDNA, 30 µg of total RNA was heat-denatured at 65°C for 5 min with 1.4 µg of oligo(dT)12−18, 4 mm dATP, 4 mm dGTP and 4 mm dTTP in 5-µl aqueous solution. Four microliters of 5 X SuperScriptII buffer (Gibco BRL, Gaithersburg, MD, USA), 2 µl of 0.1 m DTT and 8 µl of [α-33P] dCTP (Amersham Pharmacia Biotech, UK, code no. AH9905) were added and incubated at 42°C for 2 min. Then, 1 µl of SuperScriptII (Gibco BRL) was added and incubated at 42°C for 50 min to synthesize 33P-labeled single-stranded cDNA. The reaction was terminated by heating at 70°C for 15 min. To remove the unincorporated dNTPs, the reaction mixture was gel-filtrated through ProbeQuant G-50 Micro Column (Amersham Pharmacia Biotech) according to manufacturer's instruction after addition of 30 µl of 150 mm STE buffer, pH 8.0 (150 mm NaCl, 10 mm Tris–HCl, pH 8.0, 1 mm EDTA). The quality of labeled cDNA was determined according to Sasaki et al. (Tokyo Institute of Technology, personal communication) as follows: 1 µl each of purified targets was spotted onto two pieces of DEAE-cellulose paper grade DE81 (Whatman, UK; 2.3 cm in diameter). One of them was washed in 5% Na2HPO4 for 1 min three times to wash out very short cDNAs. Radioactivity on washed and non-washed filters was quantified by liquid scintillation counter using ACS II Scintillation Cocktail (Amersham Pharmacia Biotech). Only when the ratio of radioactivity on washed filter to that on non-washed filter was around 100%, labeled cDNA was used as a target for hybridization.

Hybridization procedure

A JCAA (The Japanese Consortium for Arabidopsis thaliana DNA Array) macroarray, containing approximately 13 000 non-redundant ESTs (Asamizu et al., 2000) corresponding to approximately 9000 genes, was composed of four nylon membranes (8 cm × 12 cm each) named A, B, C, and D. On the membranes A and B, each EST was spotted in duplicate, while on the membranes C and D each EST was spotted in single. To analyze the data statistically, hybridization was conducted twice for the membranes A and B, and three times for the membranes C and D to obtain more than three data for each EST. For re-hybridization, the array used for control sample in the first hybridization was hybridized with treated sample, and vice versa. For root sample, hybridization was carried out once because the amounts of total RNA extracted were too small to conduct the repetitive hybridization. Then, two (for membranes A and B) or one (for membranes C and D) data were obtained for each EST.

Two membranes out of four of one array were pre-hybridized at 65°C in 7 ml of Church buffer (Church and Gilbert, 1984) without BSA (Sasaki et al., 2001) containing 20 µg of oligo(dA)18 in a hybridization bag for at least 15 min. A half of heat-denatured 33P-labeled target cDNA prepared in a single reaction was added and hybridized at 65°C for at least 16 h. Membranes were washed twice with 0.2× SSC, 0.1% SDS at 65°C for 15 min (Sasaki et al., 2001), wrapped in a plastic wrap and exposed to the imaging plate (Fuji Photo Film, Tokyo, Japan) for at least 16 h. For re-hybridization, membranes were washed at 42°C successively with 0.1 m NaOH for 30 min and 0.1× SSC, 0.5% SDS for 30 min, to remove the 33P-labeled hybridized target. Even after removal procedure of 33P-labeled target, the signals retained on the membrane only when the signals were very strong. The intensities of retained signals were at most about 1% of those of hybridization signals and could be neglected.

Data acquisition and analyses

Hybridization signals on the imaging plates were detected using Storm 830 (Amersham Pharmacia Biotech) with the resolution of 50 µm. The image data obtained were imported into the program package Array Vision for spot detection and quantification of hybridization signals. Local backgrounds calculated in the corners between individual spots were subtracted using Array Vision to obtain raw signal intensities. The median value of all raw signal intensities, except for those corresponding to λ DNA spotted as negative control, was calculated for each membrane. The data were normalized according to the following formula:

image(1)

where RAW was raw signal intensity and MED was the median value. When the value of raw signal intensity was zero, the logarithmic value could not be calculated. Then, such data were eliminated. The number of the data before elimination was four for each EST on membranes A and B, or three for that on membranes C and D. Only when the number of the data was greater than three (n ≥ 3, membranes A and B) or greater than two (n ≥ 2, membranes C and D) in both control and treated samples, the data were used for statistical analysis to identify the ESTs whose expression was significantly changed by the treatment. In case of root samples, ESTs on the membranes A and B that had two data were further analyzed statistically. ESTs on the membranes C and D were not statistically analyzed because they had only one data (see above).

To identify the clones whose expression was significantly changed by the treatment, the statistical program SAM (Excel Add-In version), developed by Tusher et al. (2001), was used. Relative signal intensities filtered through the above procedure were used for SAM analysis. Missing data points were estimated with a K-Nearest-Neighbor imputator, where K equaled to 10 (Detweiler et al., 2001). A two-class unpaired test was conducted (Furlong et al., 2001) with a false detection rate (FDR) equal to or less than 10%, or less than 2%.

For Table 4, all data except for zero data were used for calculation of the degree of induction/repression without filtering the data because the purpose of Table 4 was to clarify the tendency of expression of all ESTs corresponding to sulfur assimilatory genes, not to analyze statistically.

Acknowledgements

This research was supported in part by CREST of JST (Japan Science and Technology Corporation), and by Grants-in-Aids from the Ministry of Education, Science, Sports and Culture, Japan. A part of the experiments was conducted in The Radioisotope Research Center of Chiba University. We are grateful to Mrs Chikara Kuwata and Taneaki Tsugane in Kazusa DNA Research Institute for helpful advices on data analysis. Macroarray filters used in this study were provided by The Japanese Consortium for Arabidopsis thaliana DNA Array (JCAA) with co-operation of Kazusa DNA Research Institute.

Supplementary Material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/TPJ/TPJ1658/TPJ1658sm.htm

Table S1 Expressed sequence tags (ESTs) significantly regulated by sulfur deficiency and/or O-acetyl-l-serine (OAS)

Ancillary

Advertisement