SEARCH

SEARCH BY CITATION

Keywords:

  • alanine : glyoxylate aminotransferase;
  • serine : glyoxylate aminotransferase;
  • peroxisome;
  • photorespiration;
  • aminotransferase;
  • Arabidopsis

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

At least two glyoxylate aminotransferases are hypothesized to participate in the steps of photorespiration located in peroxisomes. Until recently, however, genes encoding these enzymes had not been identified. We describe the isolation and characterization of an alanine : glyoxylate aminotransferase (AGT1, formerly AGT) cDNA from Arabidopsis thaliana. Southern blot analysis confirmed that Arabidopsis AGT1 is encoded by a single gene. Homologs of this class IV aminotransferase are also known in other plants, animals, and methylotrophic bacteria, suggesting an ancient evolutionary origin of this enzyme. AGT1 transcripts were present in all tissues of Arabidopsis, but were most abundant in green, leafy tissues. Purified, recombinant Arabidopsis AGT1 expressed in Escherichia coli catalyzed three transamination reactions using the following amino donor : acceptor combinations: alanine : glyoxylate, serine : glyoxylate, and serine : pyruvate. AGT1 had the highest specific activity with the serine : glyoxylate transamination, and apparent Km measurements indicate that this is the preferred in vivo reaction. In vitro import experiments and subcellular fractionations localized AGT1 to peroxisomes. Sequence analysis of the photorespiratory sat mutants revealed a single nucleotide substitution in the AGT1 gene from these plants. This transition mutation is predicted to result in a proline-to-leucine substitution at residue 251 of AGT1. When this mutation was engineered into the recombinant AGT1 protein, enzymatic activity using all three donor : acceptor pairs was abolished. We conclude that Arabidopsis AGT1 is a peroxisomal photorespiratory enzyme that catalyzes transamination reactions with multiple substrates.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

Photorespiration is a metabolite salvage pathway that is necessary because of the dual activities of ribulose-1,5-bisphosphate carboxylase/oxygenase. When ribulose-1,5-bisphosphate is oxygenated by ribulose-1,5-bisphosphate carboxylase/oxygenase, phosphoglycerate and phosphoglycolate are formed. The reactions within the photorespiratory pathway convert phosphoglycolate into phosphoglycerate, with the concomitant release of CO2 and NH3. This pathway is co-ordinated between three cellular compartments: chloroplasts, peroxisomes and mitochondria. At least two aminotransferases hypothesized to be involved in photorespiration are thought to reside within peroxisomes. These aminotransferases use the 2-oxoacid glyoxylate as an amino acceptor, with various amino acids serving as donors (Husic et al., 1987; Leegood et al., 1995; Somerville and Ogren, 1980).

Studies of plant glyoxylate aminotransferases have depended on protein preparations isolated from various plants, including spinach and cucumber (Hondred et al., 1985; Ireland and Joy, 1983; Nakamura and Tolbert, 1983; Noguchi and Fujiwara, 1982; Noguchi and Hayashi, 1980; Noguchi and Hayashi, 1981; Rehfield and Tolbert, 1972; Smith, 1973; Yu et al., 1984). Chromatographic purification analyses provide evidence for at least two distinct glyoxylate aminotransferases in plant extracts. These enzymes are active with a wide range of donor : acceptor combinations. Although these aminotransferases have somewhat overlapping substrate ranges (i.e. both catalyze an alanine : glyoxylate aminotransferase (AGT) reaction), they are distinguishable by their unique activities: one has serine : glyoxylate aminotransferase (SGT) activity, the other exhibits glutamate : glyoxylate aminotransferase (GGT) activity (Nakamura and Tolbert, 1983; Rehfield and Tolbert, 1972). There has been controversy regarding the identity of these two aminotransferases, in part due to their broad and overlapping substrate specificities, and also due to the partial purity of some of these preparations.

The confusion about the number and identity of plant glyoxylate aminotransferases is complicated by the system of nomenclature. Because genes for these enzymes were not previously identified, the enzymes were assigned names based on activity rather than homology with known genes. For consistency with nomenclature from other organisms, the Arabidopsis enzyme catalyzing AGT, SGT and serine : pyruvate (SPT) transaminations will be referred to as AGT1 because it is homologous to human AGT1 (Liepman and Olsen, 1998); this enzyme has been called AGT, SGT or tryptophan aminotransferase (Hondred et al., 1985; Nakamura and Tolbert, 1983; Noguchi and Hayashi, 1980; Smith, 1973). A second glyoxylate aminotransferase, characterized by GGT and AGT activities and a lack of SGT activity, has been called AGT, GGT or glutamate : 2-oxoglutarate aminotransferase in prior publications (Nakamura and Tolbert, 1983; Noguchi and Fujiwara, 1982; Noguchi and Hayashi, 1981; Rehfield and Tolbert, 1972). Genes encoding the aminotransferase(s) responsible for these activities in plants have not been isolated.

The hypothesis that SGT activity (AGT1) is important for photorespiration is supported by mutagenesis studies, although the mutated genes were not identified (McHale et al., 1988; Murray et al., 1987; Somerville and Ogren, 1980). Mutants lacking SGT activity were characterized by a conditional lethal phenotype when grown in conditions allowing photorespiration. Under normal atmospheric conditions these mutants grew poorly, had greatly reduced rates of photosynthesis, and accumulated serine and glycine. When photorespiration was suppressed by increasing the CO2 concentration in the growth environment, the mutants displayed no apparent phenotype, indicating that SGT activity is not required outside the photorespiratory pathway. AGT and GGT activity were maintained at nearly wild-type levels in sat mutants (Somerville and Ogren, 1980), suggesting that the aminotransferases responsible for SGT and AGT/GGT activities in vivo are probably distinct. An additional link between AGT1 and photorespiration is the finding that SGT activity and a corresponding cross-reactive polypeptide are undetectable in etiolated cucumber seedlings that do not photorespire; SGT activity increases and the cross-reactive protein becomes more abundant as the cucumber seedlings become photoautotrophic and begin to photorespire (Hondred et al., 1985; Noguchi and Fujiwara, 1982).

We have isolated the AGT1 gene from Arabidopsis (Liepman and Olsen, 1998) and studied its expression pattern. The catalytic activity of recombinant AGT1, expressed and purified from Escherichia coli, was examined free of additional plant proteins. AGT1 was localized exclusively to peroxisomes in plants. We demonstrated that air-sensitive sat mutant Arabidopsis plants contain a mutated AGT1 gene, and that when introduced into the recombinant protein this mutation abolishes all activity. These findings implicate AGT1 as a key peroxisomal enzyme in the photorespiratory pathway.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

Identification and sequence analysis of Arabidopsis AGT1

Sequence analysis of the Arabidopsis EST 35A11T7 revealed that it contains the entire coding sequence of AGT1 plus 5′ and 3′ UTR regions (formerly AGT;Liepman and Olsen, 1998). The 1441 bp cDNA contains a 1203 bp ORF, predicted to encode a polypeptide of 401 amino acids with a calculated molecular mass of 44 206 Da. A conserved variant of the type I peroxisomal targeting signal, ser-arg-ile (highlighted in Figure 1), resides at the carboxyl terminus of AGT1 (Olsen, 1998). A recent database search using the full-length protein sequence identified an AGT1 homolog from the lily Fritillaria agrestis with 85% sequence identity. Based on its sequence, the putative function of this protein was suggested to be SGT (unpublished results; Accession No. AF039000). Homologs of AGT1 with known function were found in the methylotrophic bacteria Hyphomicrobium methylovorum (Hagashita et al., 1996) and Methylobacterium extorquens (Chistoserdova and Lidstrom, 1994), as well as many mammals, including humans. The H. methylovorum homolog shares 51% sequence identity and encodes an aminotransferase specific for the SGT reaction (Hagashita et al., 1996; Izumi et al., 1990). Human AGT1 is 33% identical to ArabidopsisAGT1; in addition to an AGT reaction, human AGT1 also catalyzes a serine : pyruvate aminotransferase (SPT) reaction (Watts, 1992). Partially sequenced cDNAs from other plants, including tomato, rice, soybean, the common ice plant (Mesembryanthemum crystallinum), and the moss Physcomitrella patens, also contained regions with a high degree of similarity to Arabidopsis AGT1. It should be noted that AGT1 is distinct from alanine aminotransferase (EC 2.6.1.2) (Muench and Good, 1994; Son and Sugiyama, 1992), as demonstrated by a very low degree of sequence similarity.

image

Figure 1. Alignment of Arabidopsis AGT1 with other group IV aminotransferases.

Arabidopsis AGT1 (At, Accession No. AF063901) was aligned with a putative SGT from the lily Fritillaria agrestis (Fa, Accession No. AF039000); SGT from a methylotrophic bacterium Hyphomicrobium methylovorum (Hm, Accession No. D86125); and AGT1 from Homo sapiens (Hs, Accession no. 178273) to produce a consensus sequence (cons) using the clustalw algorithm; minor adjustments were introduced manually. Asterisks denote residues identical in all four sequences; dots indicate residues conserved in at least three of the sequences. The four residues that best conform to the aminotransferase consensus sequence are shaded in grey. Putative type 1 peroxisomal targeting sequences are boxed in black at the carboxyl terminus of the Arabidopsis, lily and human proteins. The black triangle above proline residue 251 of ArabidopsisAGT1 marks the position of the proline-to-leucine amino acid substitution in the sat mutant CS51.

Download figure to PowerPoint

Southern blot analysis indicated that a single gene encodes AGT1 in Arabidopsis (Supplementary material, Figure S-1). A BAC clone (AC007209; Lin et al., 1999), representing a region of Arabidopsis chromosome II and containing the AGT1 genomic sequence, was also identified through database searching. Results from the Southern blot were confirmed by sequence analysis of the chromosome II region containing the AGT1 gene; the size and number of hybridizing bands observed by Southern blot corresponded almost exactly to that predicted by a computer generated restriction map of that chromosomal region. Alignment of the cDNA and genomic sequences of AGT1 demonstrated that this gene contains five exons and four introns, and the transcribed region is predicted to span ≈2000 bp.

Currently, aminotransferases are divided into four subgroups based on sequence homology; Arabidopsis AGT1 most closely resembles members of group IV (Mehta et al., 1993). Examination of a comprehensive alignment of more than 50 aminotransferase sequences, including members of each subgroup, led to the discovery of four invariant residues common to all aminotransferases: gly197; asp/glu222; lys258; arg386 (Mehta et al., 1993). The absolute positions of these four residues vary among aminotransferases, but the relative spacing is conserved. The functions of these conserved residues have been defined by mapping their positions on the crystal structure of aspartate aminotransferase (Kirsch et al., 1984; McPhalen et al., 1992). In this enzyme, gly197 is located within a turn at a domain interface; asp/glu222 participates in a salt bridge to N1 of its cofactor pyridoxal 5-phosphate; lys258 forms a Schiff base with pyridoxal 5-phosphate; and arg386 is involved in a salt bridge with the α-carboxylic acid group of the substrate (Mehta et al., 1993). An alignment of Arabidopsis AGT1 with other group IV aminotransferases is shown in Figure 1. Residues gly151, asp175, lys201 and arg347 of Arabidopsis AGT1 (shaded in Figure 1) closely conform to the invariant residues characteristic of all aminotransferases. Alternatively, residues gly118, glu147, lys217 and arg347 also loosely fit the pattern, and may represent the conserved residues in Arabidopsis AGT1.

Expression of AGT1

The mRNA expression pattern of AGT1 was examined by RNA gel-blot analysis (Figure 2). AGT1 cDNA, identical to that used for Southern analysis, was used to probe filters containing total RNA isolated from whole Arabidopsis plants, cotyledons of young seedlings, aerial tissues (stems, cauline leaves, meristems, flowers), leaves, whole flowers, siliques and root tips. A transcript of ≈1.5 kb was present in each tissue type (Figure 2a) except root tips, consistent with a predicted role in photorespiration. The length of the observed AGT1 transcript is approximately equivalent to the length of the AGT1 cDNA. To control for potential loading differences, filters were also probed with the 18S rDNA fragment, and approximately equal amounts of RNA were present in each lane (Figure 2a). Quantification of the AGT1 mRNA level normalized to 18S rRNA expression in above-ground tissues of Arabidopsis is shown in Figure 2(b). AGT1 expression was highest in green leafy tissue where the transcript was about twice as abundant as it was in flowers and siliques; expression in flowers and siliques may be limited to their green portions.

image

Figure 2. Analysis of AGT1 mRNA accumulation in Arabidopsis.

Total RNA (≈5 µg) from the indicated Arabidopsis tissues was separated by electrophoresis and transferred to nylon membranes as described under Experimental procedures.

(a) Autoradiographs of membranes probed first with the 32P-labeled AGT1 cDNA fragment and subsequently with a fragment of the pea 18S rDNA, to control for potential loading differences.

(b) AGT1 transcript accumulation was quantified using phosphorimaging analysis and normalized to 18S rRNA expression in above-ground tissues. A representative experiment is shown; this experiment was repeated twice.

Download figure to PowerPoint

Purification and biochemical characterization of recombinant Arabidopsis AGT1

To study the activity of the enzyme encoded by AGT1, the protein-coding region of the cDNA was cloned into an E. coli expression vector. When protein extracts from bacterial cultures were separated using SDS–PAGE, a prominent band with the expected molecular weight of about 45 kDa was detected only in lysates of induced cells expressing recombinant Arabidopsis AGT1 (Figure 3a, lanes 2 vs. 4). Spectrophotometric assays were performed to determine whether recombinant AGT1 retained catalytic activity. The recombinant enzyme was active, and AGT activity (Figure 3b) was clearly elevated in bacterial lysates expressing the recombinant protein compared to bacteria transformed with the empty expression vector (Figure 3c).

image

Figure 3. Expression and enzymatic characterization of recombinant Arabidopsis AGT1. The entire coding sequence of AGT1 was subcloned into an E. coli expression vector (pET-28a).

(a) Lysates of bacteria expressing pET-AGT1 (lanes 3,4) or the empty pET-28 vector (lanes 1,2) were separated by SDS–PAGE and stained with Coomassie brilliant blue. Protein was collected prior to induction with IPTG (lanes 1,3), or 3 h after induction with IPTG (lanes 2,4). The size of known molecular weight markers (in kDa) is indicated.

(b) Three aminotransferase reactions catalyzed by AGT1: AGT, SGT and SPT. Only the forward reactions are indicated; the in vivo effects of the reverse reactions are probably minimal (see Discussion).

(c) Crude lysates of bacteria expressing AGT1 or the empty pET-28 vector shown in (a) were subjected to spectrophotometric enzyme assays to measure the three transaminations shown in (b). The catalytic activity of recombinant AGT1 for the SGT reaction (6.96 × 10−5 µmol min−1 mg−1) was set as 100% relative activity.

Download figure to PowerPoint

Previously isolated plant glyoxylate aminotransferases have exhibited activity with multiple donor : acceptor combinations (Hondred et al., 1985; Noguchi and Hayashi, 1980; Rehfield and Tolbert, 1972). To determine whether the AGT1 cDNA encoded an aminotransferase active with other amino donors and acceptors, recombinant AGT1 was subjected to additional aminotransferase assays. Bacterial cell lysates with AGT activity also catalyzed SGT and SPT reactions (Figure 3b,c). Recombinant Arabidopsis AGT1 was most active with serine as amino donor and glyoxylate as amino acceptor; the SGT reaction was catalyzed at a rate of 6.96 × 10−5µmol min−1 mg−1 (total E. coli protein) by the recombinant enzyme. AGT and SPT reactions were catalyzed at ≈14% and 17% the rate of the SGT reaction, respectively (Figure 3c). In-gel aspartate aminotransferase staining assays were also carried out to ensure that the AGT1 cDNA did not encode a non-specific aminotransferase. Lysates from E. coli expressing recombinant AGT1 and empty-vector controls both stained positive for the E. coli isozyme of aspartate aminotransferase, but lacked stained bands corresponding to those observed in whole extracts of Arabidopsis plants (data not shown). This result demonstrated that AGT1 was not catalyzing non-specific transaminations.

To test whether the multiple aminotransferase activities attributed to AGT1 were catalyzed by a single protein, AGT1 was chromatographically purified to apparent homogeneity (Figure 4). Purified AGT1 had the highest specific activity with the SGT reaction; it also catalyzed AGT and SPT reactions with much lower specific activity. This difference in reaction rates with various substrates can be at least partially explained by the widely variable apparent Km values for alanine and serine (Table 1). The apparent Km for alanine of AGT1 is well above predicted physiological levels, but the apparent Km for serine is within measured physiological ranges (Riens et al., 1991; Winter et al., 1993). This suggests that AGT1 primarily catalyzes an SGT reaction in vivo.

image

Figure 4. Chromatographic purification of recombinant Arabidopsis AGT1.

Recombinant Arabidopsis AGT1 was chromatographically purified from the soluble fraction of E. coli lysates as described under Experimental procedures. Samples containing approximately equal levels of SGT activity were separated by SDS–PAGE and silver-stained. The sizes of known molecular weight markers (in kDa) are indicated. Each fraction was assayed for SGT, AGT and SPT activities. The purification factor after each purification step is indicated below the corresponding lane of the gel. The initial soluble fraction (specific activity 7.58 × 10−7 µmol min−1 mg−1 total E. coli protein) was given a relative purification factor of 1. The purification factor indicated after each additional step was calculated relative to this sample.

Download figure to PowerPoint

Table 1.  Kinetic measurements of AGT1 activity with various amino acid donors and 2-oxoacid acceptors
ReactionAmino acid Km2-oxoacid concentration2-oxoacid KmAmino acid concentration
SGT1.52 mm serine1 mm glyoxylate0.11 mm glyoxylate20 mm serine
AGT101.2 mm alanine1 mm glyoxylate0.07 mm glyoxylate200 mm alanine
SPT0.39 mm serine5 mm pyruvate1.60 mm pyruvate20 mm serine

Estimates of the molecular weight of putative plant AGTs range from 62 to 185 kDa (Hondred et al., 1985; Ireland and Joy, 1983; Noguchi and Hayashi, 1980; Paszkowski and Niedzielska, 1990). The mass of recombinant Arabidopsis AGT1 was determined by gel filtration chromatography on a Superose-12 column. AGT1 eluted from the column between the β-amylase (200 kDa) and bovine serum albumin (66 kDa) standards (Supplementary material, Figure S-2). The molecular weight of AGT1 was calculated as 92 135 ± 1156 Da. This indicated that AGT1 is active as a homodimer, because the purified enzyme has an apparent subunit molecular weight of 45 kDa analyzed by standard denaturing gel electrophoresis.

Subcellular localization of AGT1

The localization of mammalian AGT1 varies under the putative influence of dietary habit. In herbivores, AGT1 is localized to peroxisomes; in carnivores it is targeted to mitochondria; and in omnivores it is found in both organelles (Danpure, 1996). We hypothesized that AGT1 was localized to plant peroxisomes because it contains a peroxisomal targeting signal, and because its substrate glyoxylate is produced in peroxisomes.

An in vitro peroxisomal protein import assay (Brickner et al., 1997) was used to determine whether AGT1 could be transported into isolated peroxisomes. Arabidopsis AGT1 was labeled with 35S-met and incubated with isolated pumpkin glyoxysomes. Samples were next treated with proteinase K to digest those proteins free in solution or merely associated with the peroxisomal membrane; proteins shielded from protease by the peroxisomal membrane were not digested (Figure 5). As expected, AGT1 was not imported into peroxisomes at 4°C. When the peroxisomal membranes were lysed with the detergent Triton X-100 and sonication prior to protease treatment, nearly all of the radiolabeled protein was digested. The pattern of AGT1 import into peroxisomes was similar to that of the control peroxisomal protein glycolate oxidase.

image

Figure 5. In vitro import of Arabidopsis AGT1 into isolated peroxisomes.

Standard in vitro peroxisomal protein import assays were performed as described under Experimental procedures. Radiolabeled translation products, either AGT1 or the peroxisomal protein glycolate oxidase (GLO), were incubated with isolated pumpkin glyoxysomes at 25°C. Following incubation, samples were treated with proteinase K to digest proteins not protected by the glyoxysomal membrane. For negative controls, import reactions were carried out at 4°C, or Triton X-100 and sonication treatments were administered prior to the addition of protease (T/S).

(a) Import samples were resolved by SDS–PAGE followed by fluorography and autoradiography.

(b) Quantification of the import samples shown in (a), using phosphorimaging analysis. The control for each protein was set as 100% relative import; this corresponds to 8% of the radiolabeled glycolate oxidase and 8.5% of radiolabeled AGT1 originally presented to the glyoxysomes. TR, translation products; C, protease-protected protein import controls. Results shown are from a representative experiment repeated four times.

Download figure to PowerPoint

Import experiments can determine if a protein is transported into peroxisomes, but they cannot rule out in vivo localization to additional subcellular compartments. Because AGT1 is found in both mitochondria and peroxisomes in some organisms (Danpure, 1996), subcellular fractionation experiments were used to further assess the localization of AGT1 in vivo. Previous attempts to localize plant glyoxylate aminotransferases indicated that there were peroxisomal isozymes of these proteins; however, because low levels of activity co-fractionated with other organellar peaks, it was not possible to exclude the possibility that AGTs reside in multiple subcellular compartments (Noguchi and Hayashi, 1980; Noguchi and Hayashi, 1981; Rehfield and Tolbert, 1972). Organelles from Arabidopsis seedling extracts were separated by sucrose density centrifugation. Gradient fractions were then assayed for the presence of marker enzymes for peroxisomes (catalase) and mitochondria (fumarase), as well as for chlorophyll (chloroplast marker) and sucrose concentration. SGT activity, the most easily assayed activity of AGT1, was also measured in each of the gradient fractions. Figure 6 shows the results from a representative subcellular fractionation experiment: the SGT activity peak co-localizes with the catalase peak at ≈53% sucrose, confirming the peroxisomal localization of AGT1. Very little SGT activity was detected in peak mitochondrial or chloroplast fractions. SGT activity was similarly localized to peroxisomes in pumpkin cotyledons (data not shown).

image

Figure 6. Localization of AGT1 to peroxisomes by subcellular fractionation.

Extracts of 2-week-old Arabidopsis seedlings were separated by sucrose density centrifugation, and each fraction assayed for SGT activity as described under Experimental procedures. Gradient fractions were also assayed for the organellar markers catalase, fumarase and chlorophyll, as well as sucrose content. The specific activity (in µmol min−1 mg−1 total protein) of each enzyme is shown (catalase × 104; fumarase × 108; SGT × 107); chlorophyll is expressed as mg chl mg−1 protein. Fraction 1 is from the bottom of the gradient; fraction 12 is the top.

Download figure to PowerPoint

Molecular characterization of the sat mutation

About 20 years ago, Somerville and Ogren devised a genetic screen to isolate Arabidopsis plants with nuclear mutations in genes encoding photorespiratory enzymes (Somerville and Ogren, 1979; Somerville and Ogren, 1980; Somerville and Ogren, 1981). One mutant, called sat, is characterized by a conditional lethal phenotype when grown under normal atmospheric conditions. The sat mutants lack SGT activity, but maintain nearly wild-type levels of AGT and GGT activities. The molecular basis of the sat mutation was unknown prior to this study of AGT1.

Because sat mutants lack SGT activity, we hypothesized that they contain a defective AGT1 gene. To test this hypothesis we sequenced AGT1 from the mutant plants. When grown under normal atmospheric conditions, the mutant seedlings were clearly stunted at the two-cotyledon stage and appeared chlorotic compared to wild-type seedlings (Supplementary material, Figure S-3). Sequence analysis of AGT1 from the sat mutant line revealed the presence of a single mutation that is predicted to result in a proline-to-leucine substitution at amino acid position 251 of AGT1. Analysis of the predicted secondary structure of AGT1 using computer programs [prof v. 1.0; psipred (Jones, 1999); predator (Frishman and Argos, 1996; Frishman and Argos, 1997)] indicated that proline 251 is situated adjacent to an α-helix.

It is possible that the proline-to-leucine substitution could alter the conformation of AGT1 near this α-helix, resulting in a grossly misfolded, or otherwise inactive protein. To determine whether a proline-to-leucine substitution at position 251 would inactivate AGT1, site-directed mutagenesis was used to introduce the sat mutation into the AGT1 cDNA. The mutated recombinant protein, when expressed in E. coli, was the same size on SDS–PAGE as wild-type AGT1. Although the recombinant AGT1 mutant protein remained soluble, it lacked any detectable SGT, AGT or SPT activity (data not shown).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

In plants, two isozymes of AGT are thought to reside within peroxisomes (Nakamura and Tolbert, 1983; Rehfield and Tolbert, 1972), but until recently (Liepman and Olsen, 1998) genes encoding these enzymes had not been identified. These aminotransferases utilize a broad and overlapping range of amino donor : acceptor combinations and may act within various metabolic pathways, including the photorespiratory glycolate cycle and general amino acid metabolism (Husic et al., 1987; Rehfield and Tolbert, 1972; Somerville and Ogren, 1982). There has been some confusion regarding the identity, substrate specificity and subcellular localization of these enzymes in plants – probably due to heterogeneity within the protein preparations and the variety of plants used for source material by previous researchers (Hondred et al., 1985; Ireland and Joy, 1983; Nakamura and Tolbert, 1983; Noguchi and Fujiwara, 1982; Noguchi and Hayashi, 1980; Noguchi and Hayashi, 1981; Rehfield and Tolbert, 1972; Smith, 1973; Yu et al., 1984).

We have isolated AGT1 from Arabidopsis, and Southern analysis confirms that there is a single AGT1 gene in Arabidopsis (Supplementary material, Figure S-1). Arabidopsis AGT1 encodes a 45 kDa class IV aminotransferase, consistent with the subunit size reported for the corresponding enzyme from spinach, cucumber and pea (Hondred et al., 1985; Ireland and Joy, 1983; Noguchi and Hayashi, 1980). Using computer searches, AGT1 homologs in many other plants including tomato, rice, soybean, the common ice plant, the lily F. agrestis, and the moss P. patens have been found, suggesting that AGT1 is conserved among all plants.

Because AGT1 is thought to play an important photorespiratory role, high levels of gene expression were expected in green tissues. AGT1 mRNA was found in all tissues, transcript accumulation being highest in green, above-ground tissues of Arabidopsis(Figure 2). A greatly reduced level of AGT1 transcripts was observed in root tips, indicating that a low level of AGT1 may be sufficient in subterranean tissues.

Purified Arabidopsis AGT1 catalyzed at least three transamination reactions: AGT, SGT and SPT 3(Figures 3 and 4). AGT1 catalyzed the SGT reaction with the highest rate; AGT and SPT reactions were catalyzed at a much lower rate. Although most aminotransferases catalyze readily reversible reactions, proposed metabolic channeling in peroxisomes (Heupel and Heldt, 1994; Heupel et al., 1991; Reumann, 2000), and the low reverse activities reported in the literature (Ireland and Joy, 1983; Nakamura and Tolbert, 1983), suggest that these AGTs function primarily in the forward direction. The observations that sat mutant plants maintain wild-type levels of AGT and GGT activity, and that purified recombinant AGT1 has highest specific activity with the SGT reaction, suggest that in vivo AGT1 functions mainly as an SGT. The apparent Km measurements of AGT1 for serine versus alanine also support a primary role for AGT1 catalysis of an SGT reaction (Table 1). Thus the AGT and GGT activities observed in sat mutants are probably attributable to one or more additional aminotransferases.

Animals have two structurally distinct enzymes with AGT activity: AGT1 and AGT2. In vertebrates, AGT2 is consistently localized within mitochondria, while the subcellular localization of AGT1 varies according to the metabolic pathway in which it is involved (Danpure, 1996; Watts, 1992). In carnivores such as dogs and cats, AGT1 is involved primarily in gluconeogenesis and is localized to mitochondria. Herbivores such as rabbits and gorillas, and descendants of herbivores including humans, harbor AGT1 within peroxisomes where it detoxifies glyoxylate, a toxic plant metabolite. In most omnivores, such as rats, AGT1 is found in both mitochondria and peroxisomes where it participates in both gluconeogenesis and glyoxylate detoxification. In humans, the metabolic disorder primary hyperoxaluria type I, marked by build-up and excessive excretion of oxalate, is caused by mislocalization of AGT1 to mitochondria rather than to peroxisomes (Danpure, 1996). We hypothesized that AGT1 is a peroxisomal protein in plants because it contains a type I peroxisomal targeting signal and its substrate glyoxylate is abundant within peroxisomes. In vitro protein import studies demonstrated that AGT1 was imported into isolated peroxisomes (Figure 5); subcellular fractionation studies confirmed a strictly peroxisomal localization of Arabidopsis AGT1 in vivo(Figure 6).

The key role of AGT1 in the photorespiratory pathway is underscored by the air-sensitive phenotype of sat mutants (Somerville and Ogren, 1979; Somerville and Ogren, 1980; Somerville and Ogren, 1981), which contain a single amino acid substitution in AGT1. These mutants photosynthesize at greatly reduced rates and accumulate serine and glycine when grown under conditions promoting photorespiration. The sat mutants can be rescued and appear to photosynthesize normally when grown in environments containing elevated CO2, where photorespiration is reduced. This suggests that AGT1 is dispensable outside the photorespiratory pathway. The mechanism by which aminotransferase activity is abolished in the sat mutant is not clear. Since proline residues often introduce kinks into polypeptides, it may be that the proline-to-leucine substitution at residue 251 could result in an unstable or aggregated protein. Although recombinant mutant AGT1 was not detectably active when expressed in E. coli, the mutant protein remained soluble (data not shown), suggesting that the conformational defect is subtle. Studies designed to understand the molecular consequences of this amino acid substitution and to further define the substrate range of AGT1 are under way. In addition, it will be interesting to determine the function and identity of other plant peroxisomal glyoxylate aminotransferases. These aminotransferases would be expected to account for the AGT and GGT activities present in sat mutant plants.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

Primer sequences

AGT1NcoI: 5′-GGAAACCATGGACTATATGTATG-3′

AGT1SalI: 5′-GAAGAGCTCGAAAGAGAAGGAC-3′

AGT1Gen1: 5′-CGAAGGAGTCCAAAACCTCCCTC-3′

AGT1Gen2: 5′-CCCTCATACCCATATAAGTTGAAG-3′

AGT1P3: 5′-GCTTGCTCTGGATTGACCAGCAG-3′

AGT1P4: 5′-CCAGCCACTTTGTTGAGACCAAG-3′

AGT1P6: 5′-GTCCACTCCCCACTCATCCATTC-3′

AGT1P9: 5′-CAATGACATCTCTGCTGTCCGCAC-3′

AGT1Mut1: 5′-CCATACACACTTTCCATTCAACTTC-3′

AGT1Mut2: 5′-GAATGGAAAGTGTGTATGGCCAATAGG-3′

Constructs

pZL-AGT1 contains a full-length Arabidopsis AGT1 cDNA in the pZL1 vector (BRL); it was obtained from the Arabidopsis Biological Resource Center (EST 35A11T7; Newman et al., 1994). PCR with Taq polymerase (Perkin Elmer, Boston, MA, USA) was used to amplify the coding region of pZL-AGT1. The mutagenic primers AGT1NcoI and AGT1SacI were used to introduce an NcoI site at the 5′ end of the ORF and a SacI site within the 3′ UTR to facilitate further manipulations. The PCR product was ligated into pCRII vector (Invitrogen, Carlsbad, CA, USA), producing pCR-AGT1. For expression in E. coli, the 1242 bp NcoI/SacI fragment from pCR-AGT1 was subcloned into the corresponding sites of the expression vector pET-28a (Novagen, Madison, WI, USA), resulting in pET-AGT1. Site-directed mutagenesis was performed on pZL-AGT1 to introduce a proline-to-leucine substitution at residue 251 of the predicted polypeptide. First, two PCR products were generated using the AGT1NcoI/AGT1Mut2 and AGT1SacI/AGT1Mut1 primer combinations and Vent DNA polymerase (NEB, Beverly, MA, USA). Next, aliquots from these reactions were used as template for a second PCR reaction with Vent polymerase and the AGT1NcoI/AGT1SacI primer pair. This PCR product was gel purified, ligated into the PCRII vector (yielding pCR-AGT1P251L) and subcloned into pET-28a in the same manner as the wild-type construct. Sequence analysis of the resulting construct (pET-AGT1P251L) indicated that only the intended mutation was introduced by these procedures. pP055 contains a fragment of the 18S rDNA from pea; it was generated in the laboratory of Dr Xing-Wang Deng at Yale University (McNellis et al., 1996) and donated by Dr Keiko Torii (University of Washington).

Nucleic acid hybridization

DNA fragments to be used as probes were labeled with 32P-dCTP (3000 Ci mmol−1, Amersham, Piscataway, NJ, USA) using a random-hexamer labeling kit (Promega, Madison, WI, USA). Unincorporated nucleotides were removed by size-exclusion chromatography through a column packed with fine Sephadex G-50, equilibrated with 10 mm Tris pH 8, 50 mm EDTA. The AGT1 probe corresponds to the 1242 bp NcoI/SacI fragment of pCR-AGT1 containing the full coding sequence of AGT1. A 1 kb EcoRI/BamHI fragment of pP055 was labeled to produce the 18S rDNA probe.

Genomic DNA was extracted from 1 g Arabidopsis (ecotype C24) leaf tissue (Dellaporta, 1994), and 10 µg genomic DNA subjected to standard Southern blot protocols (Sambrook et al., 1989). Blots were exposed to film for 10 days at −70°C.

RNA was extracted from 150 mg of various Arabidopsis tissues using a plant RNA collection system (Qiagen, Valencia, CA, USA). Root-tip RNA was generously provided by Dr Myeong-Min Lee (University of Michigan). Approximately 5 µg total RNA from each sample was subjected to standard Northern blot techniques (Sambrook et al., 1989). Blots probed with AGT1 were exposed to film for 2 days. Blots probed with 18S rDNA were exposed to film for 25 min. Band intensities were quantified by phosphorimaging analysis using a GS-363 Molecular Imager (Bio-Rad, Hercules, CA, USA).

Cloning and sequencing of AGT1 genomic fragments from wild-type and sat mutant Arabidopsis

DNA was collected from the cotyledons of sat mutant seedlings using a DNA micro-extraction technique (Li and Chory, 1998). Genomic fragments of the AGT1 gene from sat mutant seedlings and wild-type plants (ecotype Col-0) were amplified by PCR with Taq polymerase and various primer combinations (AGT1Gen1/AGT1P4, AGT1Gen1/AGT1P6, AGT1Gen2/AGT1P3, AGT1Gen2/AGT1P9). After recovering PCR products from low-melt agarose gels, fragments were cloned into the pCR-II vector (pCR-II TOPO, Invitrogen). Plasmids from clones carrying genomic fragments were isolated using standard methods (Sambrook et al., 1989), and the inserts were fully sequenced with an automated sequencer (ABI Prism 310, Perkin Elmer). To account for potential mutations introduced during amplification and cloning of genomic fragments, each region was amplified, cloned and sequenced at least twice. A single nucleotide mutation at position 751 (relative to the translation initiation site) was observed in multiple clones.

AGT1 expression in E. coli

For aminotransferase assays of bacterial lysates, E. coli BL-21 cells were transformed with pET AGT1 and grown at 37°C in 50 ml Luria Bertani (LB) super broth (32 g tryptone, 20 g yeast extract, 5 g NaCl l−1) with kanamycin selection to an OD600≈0.5 before induction with 450 µm isopropyl-beta-d-thiogalactopyranoside (IPTG). Following the addition of IPTG, cultures were shaken at 25°C. Three hours after induction, 2.5 ml aliquots were removed from each of the 50 ml cultures and the cells in these aliquots were pelleted by centrifugation at 4°C for 10 min at 10 000 g in a microcentrifuge. Bacterial cell pellets were resuspended in 200 µl lysis buffer consisting of 50 mm Tris pH 8, 10 mm EDTA, 10 mm NaCl, 10% glycerol, 0.1% β-mercaptoethanol, 1 mm phenyl methyl sulfonyl fluoride (PMSF). Cells were then mechanically lysed by repeated freeze–thaw cycles using a dry ice/ethanol bath. Bacterial lysates were quick frozen in dry ice/ethanol and stored at −70°C. Inductions of E. coli BL-21 cells expressing pET-AGT1P251L were carried out the same way, except that cells were resuspended at 10 OD600 per ml lysis buffer prior to freeze–thaw lysis.

Enzyme assays

AGT, SGT and SPT activities were all measured spectrophotometrically with a Shimadzu UV160U spectrophotometer, as described (Rehfield and Tolbert, 1972). AGT activity was measured at 340 nm by coupling the reduction of pyruvate produced by AGT1 to the oxidation of NADH, catalyzed by excess lactate dehydrogenase. To assay for AGT activity, 10 µl sample was added to a reaction mixture containing 70 mm Hepes pH 7, 0.17 mm NADH, 0.1 mm pyridoxal 5-phosphate, 1 mm glyoxylate, and 0.03 units lactate dehydrogenase in a final volume of 1 ml. SGT activity was measured at 340 nm by linking the reduction of hydroxypyruvate produced by AGT1 to the oxidation of NADH, catalyzed by excess hydroxypyruvate reductase. Reaction conditions were identical to the AGT assay except that 20 mm serine was substituted for alanine, and 0.05 units of hydroxypyruvate reductase was substituted for lactate dehydrogenase. SPT activity was measured at 340 nm by linking the reduction of hydroxypyruvate produced by AGT1 to the oxidation of NADH, catalyzed by excess hydroxypyruvate reductase. Reaction conditions were identical to the SGT assay, except that 3 mm pyruvate was substituted for glyoxylate, and 0.05 units of hydroxypyruvate reductase was substituted for lactate dehydrogenase. For each assay, the reaction was started by adding amino acid to a final concentration of 20 mm; water was added instead of amino acid to the reference cuvette. For kinetics measurements, substrate concentrations were varied over a range of 0.1–200 mm for amino acids and 0.025–10 mm for 2-oxoacids. Calculations of Km values were performed using non-linear regression software (prism program from GraphPad).

Protein concentration was measured using the BCA protein assay system (Pierce, Rockford, IL, USA). Specific activity of AGT1 for each aminotransferase reaction is expressed in µmol min−1 mg−1 and was calculated using the extinction coefficient for NADH of 6200 cm−1 m−1.

Aspartate aminotransferase activity was assayed using an in-gel staining technique (Gebhardt et al., 1998). Samples were resolved on native polyacrylamide gels that were stained by incubation in staining solution [0.1 m sodium phosphate pH 7.3, 13 mm aspartic acid, 13 mmα-ketoglutaric acid, 5 mm fast blue BB salt (catalog no. F-0250, Sigma-Aldrich Co., St. Louis, MO, USA), and 40 mm pyridoxal 5-phosphate] at 37°C for 15 min in the dark, then rinsed with water and dried.

Chromatographic purification of AGT1

Escherichia coli BL-21 cells containing the pET-AGT1 plasmid were grown at 37°C in 1.5 l LB super broth as described above. Following cell harvest, bacterial pellets were combined and resuspended in 25 ml lysis buffer, then lysed by passing them twice through a French pressure cell. Cell lysates were centrifuged at 4°C twice for 15 min at ≈50 000 g (in a Sorvall F28/50 rotor) to remove insoluble debris. Aliquots of the resulting supernatant, representing the crude soluble fraction (specific activity 7.58 × 10−7 µmol min−1 mg−1 total E. coli protein) were frozen in liquid nitrogen and stored at −70°C.

The crude soluble lysate (4 ml) was applied to a drip column packed with 10 ml DEAE cellulose resin (DE53, Whatman, Clifton, NJ, USA). AGT1 eluted with buffer A (50 mm Tris pH 8, 10% glycerol). Fractions containing AGT1 were pooled and applied to a hydroxylapatite (Bio-Rad) column connected to a fast-performance liquid chromatography (FPLC) system (Pharmacia, Piscataway, NJ, USA). AGT1 was eluted from the hydroxylapatite column by running a linear gradient of buffer A to buffer B (50 mm Tris pH 8, 400 mm sodium phosphate pH 8, 10% glycerol). Fractions containing AGT1 were again pooled and dialyzed against buffer A overnight at 4°C. AGT1 was next applied to a Q-sepharose Fast-flow (FFQ, Pharmacia) FPLC column; it was eluted from the FFQ column by running a linear gradient of buffer A to buffer C (50 mm Tris pH 8, 1 m KCl, 10% glycerol). Pooled fractions containing AGT1 were dialyzed against buffer D (50 mm MES pH 6, 10% glycerol) overnight at 4°C. The dialyzed preparation was applied to an SP-sepharose FF (Pharmacia) column, and AGT1 was eluted from this column by running a linear gradient of buffer D to buffer E (50 mm MES pH 6, 1 m NaCl, 10% glycerol). Following each chromatographic step, fractions were analyzed by SDS–PAGE and/or enzyme activity assays.

Gel filtration of AGT1

To determine the molecular weight of AGT1, ≈200 µg AGT1 (>90% pure, approximated by protein staining) in buffer A was mixed with 250 µg each of the following molecular weight standards: β-amylase (200 kDa), bovine serum albumin (66 kDa), carbonic anhydrase (29 kDa), cytochrome c (12.4 kDa), and aprotinin (6.5 kDa). This mixture was applied to an FPLC column containing Superose 12 (Pharmacia) resin. Proteins were eluted with buffer A. To locate protein peaks, 15 µl aliquots from each fraction were resolved by SDS–PAGE, and gels were silver-stained. A standard curve was constructed by plotting the logMW of the molecular weight standards against the ratio of their elution volume : void volume. This line was used to estimate the molecular weight of AGT1.

In vitro peroxisomal protein import assay

pZL-AGT1 was linearized with NotI prior to in vitro transcription with T7 RNA polymerase (Promega). mRNA synthesized in vitro was added to a cell-free wheatgerm lysate translation system containing [35S]methionine (43.5 Tbq mmol−1; Amersham). To measure translation efficiency, translation products were acid-precipitated onto glass fiber filters, washed with ethanol, and quantified with a liquid scintillation counter (Model LS-6800, Beckman, Fullerton, CA, USA).

The import of Arabidopsis AGT1 and spinach glycolate oxidase was assayed as described previously (Brickner et al., 1997) using 50 µg ml−1 proteinase K for AGT1 import reactions and 10 µg ml−1 proteinase K for glycolate oxidase import reactions. Approximately 8% of the radiolabeled glycolate oxidase and 8.5% of radiolabeled AGT1 was protected by the glyoxysomal membrane (imported) in the control import reactions.

Subcellular fractionation

Arabidopsis seedlings (ecotype C24) were grown on germination media in continuous light for 2 weeks (Olsen et al., 1993). Green seedlings were harvested and ground in a cold Waring blender with 1.5 ml grinding buffer [150 mm Hepes–KOH pH 7.5; 10 mm KCl; 1 mm MgCl2; 1 mm EDTA; 20% (w/v) sucrose; 1 mg ml−1 BSA; 5 mm DTT] g−1 FW. All subsequent steps were performed at 4°C. The resulting homogenate was centrifuged in a Sorvall HB6 (swinging bucket) rotor at 2600 g for 15 min. The supernatant was then layered onto a sucrose gradient containing 1 ml 30% sucrose, 2 ml 42% sucrose, 3 ml 45% sucrose, 2 ml 57% sucrose, and 3 ml 60% sucrose. Each sucrose solution (w/v) was prepared in 45 mm Hepes/KOH, 1 mm EDTA pH 7.5. The gradient was centrifuged in a Beckman SW40Ti rotor at 166 500 g for 4.5 h without braking. 1 ml fractions were collected from the bottom of the gradient and stored at −20°C.

Gradient fractions were analyzed for sucrose content using a Milton Roy refractometer, and for protein content using the BCA assay (Pierce). Catalase, fumarase and chlorophyll were each measured as described (Aebi, 1984; Arnon, 1949; Hatch, 1978). SGT was assayed as described above except that 0.1% Triton X-100 was included for maximal access to the enzyme (Reumann, 2000). Gradient samples were prepared with the assistance of Tanya Johnson (Olsen Laboratory, Department of Biology, University of Michigan).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

We thank Dr Wendy Crookes, Tanya Johnson and Dr Eran Pichersky for insightful discussions and critical review of this manuscript. We also thank Drs Steven Clark, Charles Yocum, Marianne Laporte, Keiko Torii and Victoria Blanc, as well as Jeannine Ross, Melissa Andrzejak and Niki Pham, for excellent technical comments and assistance. Seeds of sat mutant plants were generously supplied by Dr Chris Somerville. A.H.L. was supported, in part, by a fellowship from the University of Michigan Cellular Biotechnology Training Program (NIH-GM08353). Research in the lab of L.J.O. was funded by the USDA.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

GenBank accession number AF063901.

Supplementary material

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

Three additional figures (Figures S-1, S-2 and S-3) can be found in the online version of this paper at http://www.blackwell-synergy.com/tpj

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supplementary material
  10. Supporting Information

Figure S1. Southern blot analysis of AGT1 in Arabidopsis. (a) Schematic diagram of the region of Arabidopsis chromosome II containing the AGT1 gene. Introns are depicted as lines and exons as boxes. The shaded areas of exons correspond to the portion of the AGT1 cDNA used as a probe for DNA and RNA gel-blot analyses. The locations of predicted restriction sites are indicated. (b) Approximately 10 ?g genomic Arabidopsis (ecotype C24) DNA was digested with the indicated enzymes, separated by electrophoresis, and transferred to a nylon membrane using standard procedures. The membrane was probed with a 32P-labeled fragment of the AGT1 cDNA. The sizes of known molecular weight markers (in kb) are indicated next to the autoradiograph. Figure S2. Gel filtration analysis of recombinant Arabidopsis AGT1. Superose-12 gel filtration chromatography was used to determine the molecular mass of native AGT1. A molecular weight standard calibration curve was produced by graphing the logMW values of known molecular weight standards (b-amylase, 200 kDa; bovine serum albumin, 66 kDa; carbonic anhydrase, 29 kDa; cytochrome c, 12.4 kDa; aprotinin, 6.5 kDa) versus their elution volume to void volume (Ve/Vo) ratios. The formula of the calibration curve was logMW = 5.8478 ? 2.7401(Ve/Vo); R2 = 0.991. AGT1 has a Ve/Vo ratio of 1.4167, yielding a calculated mass of 92 449 Da. Figure S3. Arabidopsis mutant containing a proline-to-leucine substitution at residue 251 of AGT1 displays an air-sensitive phenotype. Arabidopsis seedlings were grown for 14 days under normal environmental conditions (see Experimental procedures). Both seedlings are shown at the same magnification; the mutant is ≈4 mm in diameter. (a) sat mutant seedling; (b) wild-type seedling.

FilenameFormatSizeDescription
TPJ_961_sm_f1.gif59KSupporting info item
TPJ_961_sm_f2.gif19KSupporting info item
TPJ_961_sm_f3.gif109KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.