Regulation of Serine Biosynthesis in Arabidopsis

In plants, Ser is synthesized through a couple of pathways. 3-Phosphoglycerate dehydrogenase (PGDH), the first enzyme that is involved in the phosphorylated pathway of Ser biosynthesis, is responsible for the oxidation of 3-phosphoglycerate to phosphohydroxypyruvate. Here we report the first molecular cloning and characterization of PGDH from Arabidopsis thaliana. Sequence analysis of cDNA and a genomic clone revealed that the PGDH gene is composed of three exons, encoding a 623-amino acid polypeptide (66,453 Da). The deduced protein, containing three of the most conserved regions in the NAD-dependent 2-hydroxyacid dehydrogenase family, has 38–39% identity to its animal and bacterial counterparts. The presence of an N-terminal signal sequence for translocation into plastids was confirmed by particle-gun bombardment experiments using green fluorescence protein as a reporter protein for subcellular localization. Southern hybridization analysis and restriction fragment length polymorphism mapping indicated that PGDH is a single-copy gene that is mapped to the upper arm of chromosome 1. Northern hybridization analysis indicated preferential expression of PGDH mRNA in root tissues of light-grown plants, suggesting that the phosphorylated pathway of Ser biosynthesis plays an important role in supplying Ser to non-photosynthetic tissues. The recombinant enzyme overproduced in Escherichia coli displayed hyperbolic kinetics with respect to 3-phosphoglycerate and NAD+.

3-Phosphoglycerate dehydrogenase (PGDH 1 ; EC 1.1.1.95), the first enzyme in the Ser biosynthetic pathway from 3-phosphoglycerate (3-PGA), catalyzes the oxidation of 3-PGA to form phosphohydroxypyruvate by utilizing NAD ϩ as a cofactor. Phosphohydroxypyruvate is subsequently transaminased by phosphoserine aminotransferase to yield phosphoserine. In the final step, dephosphorylation of phosphoserine to Ser is performed by phosphoserine phosphatase (Fig. 1). The molecular cloning and biochemical characterization of PGDH have been reported for a variety of bacterial (1)(2)(3) and animal (4 -6) sources. In higher plants, the biochemical characterization of PGDH enzyme preparation has been carried out in pea (7) and spinach (8); however, no investigation on molecular cloning and characterization was reported.
Ser can be formed by more than two pathways in higher plants (9,10). The photorespiratory pathway (11) of Ser biosynthesis via glyoxylate and Gly is the major route of Ser formation in photosynthetic tissues under light conditions. The glycine decarboxylase multienzyme complex (GDC), along with the enzyme serine hydroxymethyltransferase, is responsible for the respiratory conversion of Gly to Ser (12,13). The cDNAs encoding the four component enzymes of GDC (14 -17) and serine hydroxymethyltransferase (18,19) from plants have been cloned and characterized. The other two Ser biosynthetic pathways from 3-PGA via either phosphorylated or non-phosphorylated intermediates are proposed to be important Ser sources in non-photosynthetic tissues or under dark conditions, depending on the type of tissues involved (20) (Fig. 1). For these pathways, 3-PGA is supplied by glycolysis or the pentose phosphate pathway.
The serA gene, which encodes PGDH, has been cloned and characterized in Escherichia coli (1,21,22). E. coli PGDH is a tetramer of identical subunits, each consisting of three domains for nucleotide binding, substrate binding, and regulatory function. PGDHs from pea (7), E. coli (23,24), and Bacillus subtilis (25) are reported to be feedback-inhibited by Ser, whereas the enzymes from spinach (8) and animals (6,26) do not exhibit similar feedback regulation. However, PGDH in rat livers is regulated at the transcriptional level (6). In this paper, we describe, for the first time, cDNA and genomic cloning, biochemical characterization, and expression of PGDH from a higher plant, Arabidopsis thaliana.

EXPERIMENTAL PROCEDURES
Plant Materials-A. thaliana ecotype Columbia seeds were germinated and grown on germination medium (27) agar plates under 16/8-h light and dark cycles at 22°C for 3 weeks. For the dark-treated seedlings that were used for Northern analyses, 2-week-old seedlings were wrapped in aluminum foil and subsequently grown for another week before RNA extraction was carried out.
Isolation of cDNA and Genomic Clones-cDNA library screening was carried out using a 32 P-labeled synthetic 50-mer oligonucleotide probe (5Ј-GTTACTGTGACACCTCATCTTGGAGCTAGCACAAAAGAAGCT-CAGGAAGG-3Ј) that was based on the deposited sequence of the cDNA insert of the Arabidopsis expressed sequence tag FAFH01 (GenBank TM accession number ATTS3047). Approximately 2.5 ϫ 10 5 plaques from the gt11 cDNA library constructed from A. thaliana whole plants were screened. For genomic cloning, the cDNA clone CPGDH-5 was used to screen ϳ4 ϫ 10 5 amplified plaques from the Arabidopsis genomic EMBL3 SP6/T7 library (CLONTECH).
DNA Subcloning and Sequencing-The cDNA and genomic DNA inserts of the isolated clones were subcloned into appropriate cloning sites of pBluescript II(SK Ϫ ) (Stratagene). Sequencing of full-length DNA was carried out on both strands using a series of overlapping exonuclease III-digested clones created with the Exo/Mung deletion kit (Stratagene). Autosequencing was conducted by the dideoxy chain termination method with Thermo Sequenase (Amersham Pharmacia Biotech) using a Shimadzu DNA sequencer (Model DSQ1000).
Nucleic Acid Preparation and Blot Analyses-Genomic DNA was extracted from the leaves of 3-week-old seedlings as described (29). For Southern analysis, ϳ5 g of genomic DNA was digested with restriction enzymes, separated by electrophoresis through a 0.8% (w/v) agarose gel, and transferred to a Hybond N ϩ membrane. Isolation of total RNA was performed by a modified guanidine HCl method as described (28) from the leaves and roots of 3-week-old seedlings. About 10 g of total RNA was separated under denaturing conditions on a 1.2% (w/v) agarose gel containing formaldehyde and transferred to a Hybond N ϩ membrane.
DNA and RNA blots were probed with a 32 P-labeled probe synthesized from the cDNA clone CPGDH-5. To investigate the mRNA expression levels of H-protein (a subunit of GDC) and serine hydroxymethyltransferase, 32 P-labeled probes synthesized from cDNA inserts of Arabidopsis expressed sequence tag clones 200K16T7 and 111M16T7, respectively, were used. To verify equivalent loadings of RNA on blots, membranes were probed with a 32 P-labeled rice rDNA (pRR217) (30). Relative values of mRNA were calculated based on the hybridization intensities of specific signals on the blots quantified by a Fuji BAS-2000 image analyzer.
Restriction fragment length polymorphism mapping was carried out with 30 recombinant inbred lines (31). Hybridization and washing were carried out as described above, except that the final washing was performed in 0.5ϫ SSPE and 0.1% SDS for 10 min at 65°C. The 32 P-labeled genomic clone GPGDH-17 was used as a probe for hybridization. The map distance was kindly calculated by Dr. M. Arnold (Nottingham Arabidopsis Stock Center), based on the restriction fragment length polymorphism profiles generated by EcoRV.
Overexpression of Recombinant Enzyme-A general method of DNA engineering was followed as described (28). NcoI sites were created on both ends of the coding region by polymerase chain reaction engineering using the synthetic primers 5Ј-CTACACCATGGCATTTTCATCTCG-3Ј and 5Ј-CATGACCATGGATAAAACACCTT-3Ј. The engineered DNA fragment was inserted into the NcoI site of pET32a(ϩ) (Novagen), in which the cDNA was placed under a strong 10 promoter in both the sense and antisense orientations. The plasmids were then introduced into E. coli AD494(DE3) pLysS, in which the gene for lysogenic T7 RNA polymerase under the lacUV5 promoter is induced by isopropyl-1-thio-␤-D-galactopyranoside. Transformed E. coli cells were precultured in LB medium (28) supplemented with carbenicillin (100 mg/liter) at 37°C overnight and then were added to 200 ml of fresh culture medium and further cultured for 4 h. After adding isopropyl-1-thio-␤-D-galactopyranoside (1 mM) to induce gene expression, incubation was continued for another 4 h.
Enzyme Assay-Crude extracts were prepared from 50-ml stationary-phase cultures grown in LB liquid medium. Cells were harvested; washed in 20 ml of 200 mM Tris-HCl (pH 7.5), 200 mM KCl, 1 mM EDTA, and 1 mM dithiothreitol; and then resuspended in the same buffer. The cells were disrupted by sonication, and extracts were centrifuged (15,000 rpm for 15 min at 4°C) to remove debris and membrane-bound proteins (32). PGDH in the supernatant was assayed in the direction of NADH oxidation at 30°C. The assay mixture contained 10 l of enzyme extract, 25 mM Hepes (pH 7.1), 100 M NADH, 400 mM KCl, and 90 M phosphohydroxypyruvate in a final volume of 1 ml (6). The reaction was started by the addition of phosphohydroxypyruvate. One unit of enzyme activity is defined as the amount that oxidizes 1 mol of NADH/min under the indicated condition. For the physiological direction regarding enzymatic reaction, PGDH was assayed as described (6) at 30°C with some modifications. The assay mixture (1 ml) contained 10 l of extract, 200 mM Tris-HCl (pH 9.0), 25 mM EDTA, 5 mM 3-PGA, 2.5 mM dithiothreitol, and 0.5 mM NAD ϩ . One unit of enzyme activity is defined as the amount that reduces 1 mol of NAD ϩ /min under the indicated condition.
These plasmids were used for subsequent particle-gun bombardment. Particle-gun bombardment was carried out using the Helios Gene-Gun system (Bio-Rad) following the standard protocol provided by the supplier. A microcarrier loading quantity of 0.5 mg of gold/target and a DNA loading ratio of 2 g of DNA/mg of gold were chosen, and 3-week-old Arabidopsis seedlings were bombarded at a pressure of 100 p.s.i. Plates were incubated for 20 h under illumination conditions at 22°C after bombardment. Signals from individual leaves were viewed with an Olympus fluorescent microscope (BX50-FLA) using a Chroma dual band filter, FITC, and TRITC (Olympus Corp.), which provide excitations at 475-490 and 545-565 nm and emissions at 510 -530 and 585-620 nm.
Miscellaneous Techniques-SDS-polyacrylamide gel electrophoresis, protein quantitation, and primer extension were carried out as described (28

RESULTS
Arabidopsis PGDH Is Closely Related to Its Counterparts from B. subtilis, Synechocystis sp., and Mammals-Phage plaques produced from an Arabidopsis whole plant cDNA library were screened with a synthetic 50-mer oligonucleotide probe based on the sequence of A. thaliana expressed sequence tag clone FAFH01, which shows high homology to PGDH from B. subtilis. Among the four positive clones selected for further studies, CPGDH-5, which contains the largest cDNA insert (2.2 kilobase pairs), was subcloned and sequenced. Sequence analysis revealed an open reading frame of 1881 nucleotides, encoding for 623 amino acid residues. The first ATG triplet, which is 14 nucleotides away from the 5Ј-end of CPGDH-5, is designated as the translational start point because the sequence around the Met codon (AGTCATGGC) matches well with the consensus sequence for plant gene initiation codons (AACAAT-GGC) (36). This is further supported by the primer extension result that mapped the transcriptional start point at 38 bp before the translational start site. A 3Ј-untranslated region of 265 nucleotides downstream of the translational stop codon (TAA) is present in the cDNA sequence. The AATAAA polyadenylation signal is located 129 nucleotides upstream of the poly(A) tail.
The deduced amino acid sequence of Arabidopsis PGDH has been aligned with PGDHs from other organisms (Fig. 2). A phylogenetic tree (Fig. 3) indicates that Arabidopsis PGDH is closely related to the enzymes from B. subtilis, Synechocystis sp., and mammals. These proteins form a family distinct from other bacterial and yeast PGDHs. The deduced 66,453-Da protein, containing three of the most conserved regions in the NAD-dependent 2-hydroxyacid dehydrogenase family, has 38 -39% similarity to the amino acid sequences of PGDHs from other organisms (Fig. 2). The first pattern is based on a Glyrich region that probably corresponds to the NAD-binding domain, Gly-X-Gly-X 2 -Gly-X 17 -Asp (37). Two other patterns con-tain a number of conserved charged residues, some of which may play a role in the catalytic mechanism. The 623 amino acid residues of Arabidopsis PGDH, sharing the three-dimensional structure of E. coli PGDH (22), is the longest sequence among all. It differs from the rest mainly due to its C-terminal domain and N-terminal extension.
PGDH Gene of Arabidopsis Is Mapped to Chromosome 1-The cDNA clone CPGDH-5 was used to screen the Arabidopsis genomic DNA library. Among various clones, GPGDH-17 was selected for further studies. The 7.1-kilobase pair fragment that covers the PGDH structural gene and the 5Ј-and 3Ј-

3-Phosphoglycerate Dehydrogenase from Arabidopsis
flanking regions are presented in Fig. 4A. Analysis of genomic and cDNA sequences revealed the presence of two introns. All of the exon/intron junctions had the consensus GT/AG splice donor and acceptor sites. The sequences for both the coding regions and the 3Ј-and 5Ј-untranslated regions of the cDNA clone are identical to those for the genomic clone, demonstrating that this gene is actively transcribed. The transcriptional start site of the genomic clone was determined using primer extension analysis. A single major transcriptional start point was confirmed to be located 38 bp before the translational start site. An AT-rich sequence is located at Ϫ27 to Ϫ34 bp with respect to the transcriptional start site (ϩ1), and a potential CAAT sequence is located at Ϫ79 to Ϫ82 bp (Fig. 4B).
Using the coding sequence of PGDH as a probe, the Southern blot results (Fig. 5) suggest that Arabidopsis PGDH is a singlecopy gene. Upon digestion with BglII, EcoRI, EcoRV, SacI, and XbaI, several bands were observed due to the presence of several restriction sites for these endonucleases in the genomic sequence. The Arabidopsis PGDH gene was mapped to the upper arm of chromosome 1 between the markers g3786 and g3829.
Arabidopsis PGDH Is a Plastidic Protein-The 60-amino acid leader sequence exhibits the general features of a transit peptide for transportation of protein to plastid. It starts with Met-Ala; is rich in hydroxylated amino acids, Ser and Thr (19/60); is rich in small hydrophobic amino acids, Ala and Val (13/60); is essentially deficient in acidic amino acids, Asp and Glu (1/60); and has a net positive charge (pI 11.5). Prediction by the PSORT program 2 also suggested its localization in chloroplasts. The recombinant fusion protein CPGDH-GFP, containing the N-terminal 82-amino acids fused with GFP, could be detected in intact tissues after delivering the constructs into Arabidopsis leaves by particle-gun bombardment. The observed signals in the construct containing the predicted Nterminal transit peptide of PGDH from Arabidopsis were observed as green fluorescence that lighted up the chloroplasts (data not shown). They were similar to those exhibited by GFP fused with the transit peptide of the ribulose-1,5-bisphosphate carboxylase/oxygenase small subunit polypeptide of Arabidopsis (34), which was already known to be sufficient for translocation of a passenger protein to chloroplasts (33). Despite a fainter degree of signals exhibited by CPGDH-GFP compared with the positive control, the fluorescent pattern observed for CPGDH-GFP was clearly distinct from those exhibited by mitochondrial and cytosolic proteins. These results confirmed that the N-terminal sequence of Arabidopsis PGDH is sufficient for translocation of passenger protein into chloroplasts, and thus, PGDH is a plastidic protein.
Arabidopsis PGDH Can Functionally Complement E. coli serA Ϫ Mutant-The identity of the isolated cDNA clone CPGDH-5 was confirmed by successful complementation of the E. coli serine auxotroph 536 (HfrOR11 glu V42 Ϫ SerA13 T3 rr ) (1) (Fig. 6). Mutant E. coli cells were transformed with the expression plasmid pPGDH-AB14, in which the expression of PGDH is regulated by the lacZ promoter. Transformants could grow on M9 minimal medium in the absence of Ser, whereas pTV118N-transformed E. coli 536 cells were not able to grow without supplementation of Ser, indicating the authenticity of CPGDH-5 encoding the functional PGDH. The shortened cDNA with 95 amino acids truncated at the N terminus was not able to complement the E. coli 536 mutant. This failure of complementation could be due to deletion of amino acid residues conserved among PGDHs from different organisms and thus necessary for functional expression. 2 GenomeNET Service, Osaka University, Osaka, Japan.

3-Phosphoglycerate Dehydrogenase from Arabidopsis
Biochemical Properties of Recombinant Arabidopsis PGDH Produced in E. coli-Recombinant PGDH was overproduced in E. coli AD494 cells using a pET32a(ϩ) vector system with a strong promoter. The recombinant protein was visualized on SDS-polyacrylamide gel as the expected 90-kDa protein in the insoluble fraction of crude extract as an inactive form. However, production of PGDH in the soluble fraction was too low to be visualized by SDS-polyacrylamide gel electrophoresis. Nevertheless, the soluble form of the protein exhibited PGDH activity of 0.14 Ϯ 0.01 units/mg of protein in the physiological direction, catalyzing the oxidation of 3-PGA to phosphohydroxypyruvate, and of 10.95 Ϯ 1.36 units/mg of protein in the opposite direction, reducing phosphohydroxypyruvate to 3-PGA. Enzyme activity was not detected in the cells transformed by the cDNA in the antisense orientation relative to the promoter. The construct of an insert with a 95-amino acid truncation at the N terminus could not be overexpressed in E. coli AD494 cells.
Double-reciprocal plots of the data for the initial rates demonstrated K m values of 0.35 and 0.12 mM for phosphohydroxypyruvate and NADH, respectively, at pH 7.1. The activity was inhibited by phosphohydroxypyruvate (ϳ90 M), as reported for the rat enzyme (6). This inhibition could be released by 100 -400 mM KCl. K m values for 3-PGA and NAD ϩ were 1.19 and 0.01 mM, respectively, at pH 9.0. Ser, Thr, Val, Gly, Trp, O-acetyl-L-Ser, and Cys (in the range of 5-50 mM) had no effect on the reaction rates in both orientations.
Preferential PGDH Expression in Root Tissues-The mRNA abundance of PGDH was examined in leaf and root tissues from both light-grown and dark-treated plants. The highest level of PGDH mRNA expression was observed in light-grown root tissues (Fig. 7A). It was ϳ2-3-fold higher than the mRNA expression in dark-grown root and leaf tissues. A minor amount of mRNA expression (ϳ1:15 of roots in light) was detected in the light-grown leaf tissues. The preferential expression of PGDH in root tissues of light-grown plants was in contrast with the expression pattern exhibited by H-protein (a component protein of GDC) (13) and serine hydroxymethyltransferase (18,19), which accumulated primarily in the lightgrown leaf tissues (Fig. 7, B and C). These RNA blot analyses suggested that the regulation of the PGDH gene is mainly exerted at the level of transcription or by stability of mRNA. DISCUSSION This is the first investigation on the molecular characterization of plant PGDH, a key enzyme committed to the entry step of the phosphorylated pathway of Ser biosynthesis. The isolated cDNA contains an open reading frame encoding the entire PGDH polypeptide of A. thaliana. The deduced protein with a molecular mass of 66,453 Da, sharing the three-dimensional structure of the E. coli enzyme (22), is composed of three distinct domains: a nucleotide-binding domain, a substratebinding domain, and a regulatory domain or a Ser-binding domain in each subunit of the tetrameric protein of E. coli PGDH. The main contact points between the subunits are at the level of the coenzyme-binding domains and the regulatory domains, indicating the importance of these zones for tetramerization. The deduced amino acid sequence has high similarity to eukaryotes (human and rat), but not yeast. Surprisingly, the nucleotide-and substrate-binding domains of B. subtilis PGDH exhibit more similarity to the eukaryotic enzymes than to other bacterial PGDH enzymes (E. coli and Haemophilus influenza), whereas the yeast enzyme is closer to the latter. This suggests that there are two different types of PGDH that may have evolved at its origin before diverging to eukaryotes and prokaryotes. Three of the most common regions in the NAD-dependent 2-hydroxyacid dehydrogenase family are conserved in Arabidopsis PGDH.  7. Northern analysis of RNA from different tissues of Arabidopsis seedlings. Ten g of total RNA was separated under denaturing conditions on a 1.2% (w/v) agarose gel containing formaldehyde, transferred to a Hybond N ϩ membrane, and then probed with a 32 Plabeled cDNA clone. The final washing was performed in 0.1ϫ SSPE and 0.1% SDS at 65°C for 10 min. Left panels, RNA blot results; right panels, relative mRNA abundance normalized with the rRNA expression data as a control. A shows the preferential expression of PGDH mRNA in root tissues of light-grown plants. In B and C, the preferential expression of mRNA from H-protein of GDC and serine hydroxymethyltransferase (SHMT) was observed in leaf tissues of light-grown plants, although a much lower amount of expression was also detected in dark-treated leaf tissues.

3-Phosphoglycerate Dehydrogenase from Arabidopsis
Alignment of the Arabidopsis PGDH sequence with bacterial, yeast, and mammal sequences reveals the presence of a presequence, presumably targeting the nuclear encoded protein to the chloroplasts/plastids. The essential common features of the chloroplast presequence are exhibited by the first 60 deduced amino acid residues at the N terminus. The exact cleavage site of the transit peptide was difficult to determine. Based on the multiple alignment results (Fig. 2) showing a low homology between various organisms at its first 80 amino acids, together with our complementation experiment and attempts to overexpress PGDH protein with its 95 amino acids truncated at the N terminus, the cleavage site is most probably located between 80 and 90 amino acids away from the N terminus. Even if the protein is actually processed after entering the chloroplasts, the kinetic properties may not change much. From our recent results with phosphoserine aminotransferase (38), the full-length proteins and the proteins with the transit peptide truncated exhibited essentially the same properties.
We have provided evidence for visualization of the targeting of the fusion protein of the N-terminal transit peptide and GFP to leaf chloroplasts. The weaker signals shown by CPGDH-GFP compared with the positive control of the ribulose-1,5-bisphosphate carboxylase/oxygenase transit peptide may be due to its intrinsic nature. PGDH has been detected in the proplastids of soybean (39,40), and there may be an extrachloroplastic form, too (8). However, Southern blot analysis and restriction fragment length polymorphism results confirmed that Arabidopsis PGDH is a single-copy gene that is mapped to the upper arm of chromosome 1.
We have provided molecular evidence that the mRNA of PGDH is preferentially expressed in roots, similar to that of phosphoserine aminotransferase, being the subsequent enzyme of the phosphorylated pathway (41). Conversely, the mRNA abundance of the H-protein (a subunit of GDC) and serine hydroxymethyltransferase, which both are responsible for Ser biosynthesis in the photorespiratory pathway, in leaves of light-grown plants far exceeded expression in roots or darkgrown leaves. Therefore, we suggest that the phosphorylated pathway may take over as the major route of the Ser biosynthetic pathway in both non-green tissues and green tissues in the dark when the photorespiration rate is low.
Lower enzyme activities of PGDH in the physiological direction compared with the non-physiological direction were also reported by Slaughter and Davies (7) under the given conditions. They reported that pea PGDH was inhibited by Ser, but became progressively less sensitive as the purification progressed. However, Larsson and Albertsson (8) could not find Ser or phosphoserine inhibition of the enzyme in spinach chloroplast extract. Our results concur with those of Larsson and Albertsson, i.e. Ser does not show any inhibition effect on the enzyme activities. Hence, Ser levels are unlikely modulated by the metabolite flux of the pathway from 3-PGA to Ser through enzyme activities. Although the recombinant protein produced in E. coli has been used effectively to verify gene product function, we have to be careful when making conclusions regarding regulation since few potentially post-translational events may occur in E. coli. Nevertheless, our present data indicated that the regulation of PGDH is mainly exerted at the level of transcription or by stability of PGDH mRNA.