Arabidopsis Mutants Define an in Vivo Role for Isoenzymes of Aspartate Aminotransferase in Plant Nitrogen Assimilation

Corresponding author: Gloria M. Coruzzi, New York University, Biology Department, 100 Washington Square East, 1009 Main Bldg., New York, NY 10003, coruzg01{at}mcrcr6.med.nyu.edu (E-mail).

Communicating editor: E. MEYEROWITZ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Arabidopsis contains five isoenzymes of aspartate aminotransferase (AspAT) localized to the cytosol, chloroplast, mitochondria, or peroxisomes. To define the in vivo function of individual isoenzymes, we screened for Arabidopsis mutants deficient in either of the two major isoenzymes, cytosolic AAT2 or chloroplastic AAT3, using a native gel activity assay. In a screen of 8,000 M₂ seedlings, three independent mutants deficient in cytosolic AAT2 (aat2) and two independent mutants deficient in chloroplastic AAT3 (aat3) were isolated. Mapping of aat2 and aat3 mutations and the five AspAT genes (ASP1–ASP5) established associations as follows: the mutation affecting aat2 maps with and cosegregates with ASP2, one of two expressed genes for cytosolic AspAT; the mutation affecting aat3 maps to the same location as the ASP5 gene encoding chloroplastic AspAT. Phenotypic analysis of the aat2 and aat3 mutants revealed a dramatic aspartate-related phenotype in one of the mutants deficient in cytosolic AAT2. The aat2-2 mutant displays an 80% reduction in levels of aspartate transported in the phloem of light-grown plants, and a 50% reduction in levels of asparagine transported in dark-adapted plants. These results indicate that cytosolic AAT2 is the major isoenzyme controlling aspartate synthesized for nitrogen transport in the light, and that this aspartate pool is converted to asparagine when plants are dark adapted.

MANY enzymes involved in plant metabolism exist as multiple isoenzymes, some of which are targeted to distinct subcellular compartments (WENDEL and WEEDEN 1989 ). Understanding whether these isoenzymes play overlapping or distinct roles in vivo is a question that remains open for many isoenzyme families. For some, the putative in vivo roles of individual isoenzymes have been addressed using molecular and transgenic approaches. For example, chloroplastic and cytosolic isoenzymes of glutamine synthetase (GS) are encoded by multiple genes in all higher plants studied including Arabidopsis (PETERMAN and GOODMAN 1991 ). Promoter-GUS fusions and immunocytochemistry have shown that cytosolic and chloroplastic isoenzymes of GS are each expressed in distinct cell types in several species examined, implying distinct in vivo functions (FORDE et al. 1989 ; EDWARDS et al. 1990 ; CARVALHO et al. 1992 ). As traditional biochemical analyses of isoenzymes cannot address the in vivo significance of cell-specific or subcellular compartmentation, the function of individual isoenzymes in planta has remained largely unaddressed.

An isoenzyme family that has received particular attention at the biochemical level is aspartate aminotransferase (AspAT, E.C. 2.6.1.1), which plays a key role in both nitrogen and carbon metabolism in many organisms. In plants, distinct AspAT isoenzymes have been localized to each of four subcellular compartments: the cytosol, chloroplasts, mitochondria, and peroxisomes, as shown for several plant species including Arabidopsis (LIU and HUANG 1977 ; WEEDEN and MARX 1987 ; SCHULTZ and CORUZZI 1995 ; WILKIE et al. 1995 ). These distinct AspAT isoenzymes are believed to be involved in shuttling reducing equivalents between subcellular compartments, or between cells, and to be involved in the assimilation of nitrogen into aspartate which serves as an important nitrogen donor and nitrogen-transport compound in plants (IRELAND and JOY 1985 ; GIVAN 1990 ).

To attempt to address the function of the distinct AspAT isoenzymes in plants, we initiated a molecular-genetic study of the ASP gene family and AAT isoenzymes in Arabidopsis. Using native gel assays combined with subcellular fractionation, we showed that mitochondrial AAT1 is a minor component of Arabidopsis extracts, while cytosolic AAT2 and chloroplastic AAT3 predominate in all tissues examined (leaves, roots, flowers, and cotyledons) (SCHULTZ and CORUZZI 1995 ). At the molecular level, Arabidopsis has been shown to contain five genes for AspAT (ASP1–ASP5) encoding isoenzymes localized to distinct subcellular compartments. The ASP2 and ASP4 genes each encode cytosolic isoenzymes, with ASP2 being the most highly expressed gene (especially in roots) based on the analysis of steady state mRNA levels (SCHULTZ and CORUZZI 1995 ). The ASP1 gene was predicted to encode a mitochondrial isoenzyme, while ASP3 was predicted to encode either a plastid or peroxisomal enzyme based on transit peptide sequence analysis (SCHULTZ and CORUZZI 1995 ). A fifth ASP gene was identified in Arabidopsis (pcAtAAT1, referred to herein as ASP5) and appears to encode a chloroplastic AspAT isoenzyme based on in vitro chloroplast uptake experiments (WILKIE et al. 1995 ).

To uncover the in vivo role of specific AspAT isoenzymes, we developed a screen to identify Arabidopsis mutants deficient in either of the two major isoenzymes of AspAT, cytosolic AAT2, and chloroplastic AAT3. Subsequent phenotypic analysis of the mutants was used to provide insights into the in vivo function of each isoenzyme. This mutant approach, outlined herein, has enabled us to determine that the cytosolic AAT2 isoenzyme controls the major flux of nitrogen assimilated into aspartate, which is used to transport nitrogen from sources to sinks.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant lines and growth conditions:
Plant lines used in all experiments were of the Columbia (Col) ecotype of Arabidopsis thaliana. The Landsberg (Ler) ecotype was used for mapping purposes only. Plants were grown in tissue culture or soil (as noted) in EGC growth chambers (Environmental Growth Chambers, Chagrin, OH) set on a 16-hr light (65 µE·m^-2·sec^-1)/8-hr dark cycle, unless otherwise noted. Mutagenized M₂ Arabidopsis Columbia seeds treated with ethylmethane sulfonate (EMS) or nitrosourea were kindly donated by ROBERT LAST (Boyce Thompson Institute, Cornell University). For screening, M₂ seeds were surface sterilized and germinated on Murashige and Skoog (MS) media containing 3% sucrose and 0.05% aspartate, to enable the isolation of putative aspartate auxotrophs.

Nomenclature:
Genes encoding aspartate aminotransferase isoenzymes were named ASP1–ASP4, as described previously (SCHULTZ and CORUZZI 1995 ). ASP5 refers to a fifth Arabidopsis AspAT gene described by WILKIE et al. 1995 , WILKIE et al. 1996 . The mutants deficient in cytosolic or chloroplastic AspAT isoenzymes are named aat2 and aat3, respectively, to distinguish them from the ASP genes. This nomenclature is consistent with the community standards for Arabidopsis genetics (MEINKE and KOORNNEEF 1997 ).

AspAT activity gels:
For the AspAT native gel assays, one to three leaves of M₂ seedlings were ground in 20 µl grinding buffer (50 mM TrisCl pH 7.5, 5% glycerol, 0.1% Triton X-100) and extracts were clarified by centrifugation. Supernatants were electrophoresed through nondenaturing, discontinuous PAGE mini-gels (mini protean II; Bio-Rad, Richmond, CA) and stained for AspAT activity at room temperature with gentle shaking for 15–60 min. Stain was made fresh for each gel by adding 0.05 g of fast blue BB (F0250; Sigma, St. Louis) to 50 ml of AspAT substrate solution, pH 7.4 (WENDEL and WEEDEN 1989 ). AspAT substrate solution is stable at room temperature for up to 6 months and is composed of 2.2 mM -ketoglutaric acid (K1875; Sigma), 8.6 mM L-aspartic acid (A6683; Sigma), 0.5% polyvinyl pyrrolidone-40 (PVP-40, Sigma), 1.7 mM EDTA (disodium salt), 100 mM sodium phosphate (dibasic).

Mapping the ASP1–ASP5 genes:
Restriction fragment length polymorphisms (RFLPs) or CAPS (cleaved amplified polymorphisms, KONIECNZY and AUSUBEL 1993 ) were identified between Col and Ler ecotypes of Arabidopsis for the ASP1–ASP5 genes. The enzymes used to generate the polymorphisms were as follows: for ASP1, DraI or AvaI; for ASP2, HpaII or AvaI, for ASP3, DraI or AvaI, for ASP4, TaqI and for ASP5, AciI. The gene-specific probes used to identify RFLPs for ASP1–ASP3 were described previously (SCHULTZ and CORUZZI 1995 ). The primers used to generate the CAPS marker for ASP4 were CS-13 (5' GAGAGTTGGAGCTGAG 3') and CS-51 (5' CGGCTACAAACATACGAACC 3'). The primers used to generate the PCR probe for RFLP analysis of ASP5 were BM24 (5' CAATCAATGTCGTGTGCTCC 3') and BM17 (5' TCGCATCAGCAAGATACTCG 3'). The cDNA and genomic clones representing ASP5 were described by WILKIE et al. 1995 , WILKIE et al. 1996 . ASP1, ASP2, ASP3, and ASP5 were mapped with the Lister and Dean Recombinant Inbred lines generated from the Columbia and Landsberg ecotypes (LISTER and DEAN 1993 ) using 22, 28, 23, and 20 individuals, respectively. The segregation data were analyzed and placed on the genetic map by C. LISTER (John Innes Centre, Norwich, UK) or since July 1996, by M. ANDERSON (Nottingham Arabidopsis Stock Centre, WWW server http://nasc.nott.ac.uk/). ASP4 was mapped relative to known CAPS or SSLP (simple sequence length polymorphisms) markers using 27 plants (BELL and ECKER 1994 ).

Mapping the aat2 and aat3 mutant alleles:
CAPS and SSLP analysis was used to map the genes affected in the aat2-4 and aat3-3 mutants. Mapping populations were generated for aat2-4 or aat3-3 from the following crosses, respectively: aat2-4/aat2-4 (Col) x AAT2/AAT2 (Ler) or aat3-3/aat3-3 (Col) x AAT3/AAT3 (Ler). Homozygous mutants were identified by screening approximately 120 individuals from the appropriate segregating F₂ population. Note, the aat3-3 mutant is not described here in detail because the mutation affects the electrophoretic mobility of chloroplastic AAT3 rather than causing a loss of activity (SCHULTZ 1994 ).

Genetic characterization of the aat mutants:
The individual aat mutants were outcrossed to wild-type Columbia to eliminate background mutations. aat2-1 was outcrossed over five generations, aat2-4 was outcrossed over three generations, and aat2-2, aat3-1, and aat3-2 were outcrossed over one generation. To show that the aat2-1 mutant gene segregated in a semidominant manner, three putative heterozygotes and four putative homozygous mutants from the F₂ generation (from the cross aat2-1/aat2-1 x AAT2/AAT2) were selfed. At least 10 (and up to 37) F₃ individuals were analyzed from each of the seven F₂ individuals. The following crosses were performed to test for allelism. The pollen recipient is listed first and the number of seeds obtained (and analyzed) is given after each cross; aat2-1 x aat2-2, 10 seeds; aat2-1 x aat2-4, 13 seeds; aa3-2 x aat3-1, two crosses, 11 and 13 seeds, respectively. To minimize the risk of self-fertilization the flowers chosen as pollen recipients were at a stage where the pollen on the attached anthers was not mature. All anthers were removed prior to touching the stigma with mature donor pollen. The plants chosen as pollen recipients were just starting to send up flowering bolts and the surrounding flowers (not used in the crosses) were removed, to minimize the possibility of self-pollination from a neighboring flower. Since the frequency of spontaneous outcrossing in Arabidopsis is very low (approximately 0.05%, REDEI and KONCZ 1992 ) it is very unlikely that the results of the crosses are due to spontaneous self-fertilization by a neighboring flower. To determine whether any of the aat2 mutants cosegregate with any of the ASP genes, the aat mutants (Col) were outcrossed to Landsberg and the F₂ progeny were analyzed.

Analysis of growth rate:
To compare growth rate of aat mutants and wild-type Col, aat2 or aat3 mutants and wild-type (Col) seeds were sown side-by-side in a row on MS media containing 3% sucrose. The plates were incubated vertically using a 16-hr light/8-hr dark regimen, and root length was assessed by visual inspection as an indicator of growth rate.

HPLC analysis of free amino acids in phloem exudates:
To assess the levels of amino acids transported via the phloem in leaves of wild-type and mutant plants, phloem exudates were obtained using a method reported for pea (URQUHART and JOY 1981 ) and modified for Arabidopsis as follows: single Arabidopsis rosette leaves (from soil grown plants) were cut from the plant to leave as much petiole attached to the leaf as possible. Leaves were immediately placed in a microfuge tube containing 50 µl of 20 mM EDTA pH 7.0 such that 1–2 mm of petiole was submerged. Phloem exudates were collected from either light-grown or dark-adapted plants, for two hours in the light and dark, respectively (LAM et al. 1995 ). Control exudates into water yielded minimal levels of amino acids. After 2 hr the leaves were removed from each tube and the final volume measured by pipette to account for any increase in volume due to the exudate. Samples were diluted one in three and filtered (#DDN02003NB; Micron Separations, Westborough, MA) prior to HPLC analysis. Samples were derivatized at 4° with o-phthaldialdehyde immediately prior to injection using an autosampler and then separated by reverse phase-HPLC (SCL-10A system; Shimadzu, Tokyo, Japan) on a C18 column (Supelcosil LC-18, 25 cm x 4.6 mm, 5 µm; Supelco, Bellfonte, PA) at room temperature. Amino acids were separated with a gradient of buffer A (0.1 M sodium acetate pH 7.2, 4.5% methanol, 0.5% tetrahydrofluran) starting at 27.5% and finishing with 100% buffer B (80% methanol). The gradient was determined empirically as follows: time-Buffer B(%), 0.01 min-27.5% B, 38 min-27.5% B, 39 min-33% B, 49 min-65% B, 73 min-66.3% B, 75 min-75% B, 78 min-80% B, 83 min-100% B, then hold 5 min. Flow rate was 1 ml/min. Derivatized amino acids were detected using a Perkin-Elmer LS30 fluorimeter (excitation wavelength 360 and emission wavelength 455). Amino acid standards were from Sigma. Amount of each amino acid was determined by linear comparison of five standard runs where each amino acid was present at 1000 pmol, 500 pmol, 100 pmol, 10 pmol, 5 pmol. The standards all gave linear correlation of concentration to area under the curve for each peak as determined by linear regression.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Use of a quantitative gel assay to screen for Arabidopsis mutants deficient in cytosolic AAT2 or chloroplastic AAT3:
AspAT holoenzymes present in Arabidopsis were detected using an AspAT activity stain on leaf extracts run on nondenaturing gels (SCHULTZ and CORUZZI 1995 ). AspAT activity gels show that crude leaf extracts of wild-type Arabidopsis contain two prominent AspAT isoenzymes; cytosolic AAT2 and chloroplastic AAT3 (Figure 1, lane 1). A mitochondrial form of the enzyme (AAT1) is low in abundance and rarely detected in crude extracts but is detected in preparations of partially purified mitochondria (SCHULTZ and CORUZZI 1995 ). The AAT2 and AAT3 isoenzymes are the predominant AspAT isoenzymes detected in all other tissues examined including cotyledons, roots, stems, and flowers (data not shown). The native gel assay for AspAT activity was shown to be quantitative using serial dilutions of crude leaf extracts. Cytosolic AAT2 can be detected in samples with 6% of wild-type activity remaining (Figure 1, lane 5), while chloroplastic AAT3 activity is detectable in samples with 3% of wild-type activity remaining (Figure 1, lane 6). As the AspAT native gel assay was deemed to be quantitative, it was used to screen for Arabidopsis mutants deficient in either cytosolic AAT2 or chloroplastic AAT3. In a screen of 8000 M₂ seedlings, several independent mutants deficient in either cytosolic AAT2 or in chloroplastic AAT3 were identified. Three independent loss-of-activity mutants deficient in cytosolic AAT2 (aat2-1, aat2-2, and aat2-4) were identified from separate pools of EMS mutagenized seeds. Lines homozygous for each of the three aat2 mutants contain no detectable AAT2 activity, as judged by the native gel assay (for example Figure 2A, lane 3). Thus, the aat2 mutants each contain less than 6% of wild-type AAT2 activity, as judged by quantitations of dilutions of wild-type extracts in the native gel assay (see Figure 1). For chloroplastic AAT3, two independent loss-of-activity mutants were isolated from separate pools of nitrosourea-treated seeds (aat3-1 and aat3-2). Lines homozygous for each of the two aat3 mutants contain no detectable chloroplastic AAT3 activity in native gel assays (Figure 2B, lane 3) and therefore contain less than 3% of wild-type AAT3 activity (see Figure 1).

View larger version (46K): In this window In a new window Download PPT slide   Figure 1. —A quantitative gel assay for cytosolic AAT2 and chloroplastic AAT3. An extract made from three wild-type Arabidopsis (Col) rosette leaves ground in 20 µl grinding buffer (approximately 100 mg protein) (lane 1) was serially diluted (twofold; lanes 2–6) and separated by nondenaturing PAGE and stained for AspAT activity. The top band represents cytosolic AspAT (AAT2) and the bottom band represents chloroplastic AspAT (AAT3).

View larger version (59K): In this window In a new window Download PPT slide   Figure 2. —Arabidopsis mutants deficient in cytosolic AAT2 or chloroplastic AAT3. (A) Gel phenotype of an aat2 mutant deficient in cytosolic AAT2. Representative F2 individuals from a segregating population resulting from the selfing of AAT2/aat2-2. The genotype of each F2 individual is shown above each lane; lane 1 wild type (+/+), lane 2 heterozygote (+/-), and lane 3 homozygous mutant (-/-). (B) Gel phenotype of a mutant deficient in chloroplastic AAT3. Representative F2 individuals from a segregating population resulting from the selfing of AAT3/aat3-2. The genotype of each F2 individual is shown above each lane: lane 1 wild-type (+/+), lane 2 heterozygote (+/-), and lane 3 homozygous mutant (-/-). (C) Gel phenotype of five randomly selected individuals in the F3 generation as a result of selffertilization of a putative heterozygote (AAT2/aat2-1) from the F2 generation. (D) Gel phenotype of five randomly selected individuals in the F3 generation as a result of self-fertilization of a putative heterozygote (AAT3/aat3-2) from the F2 generation. (E) Gel phenotype of F3 individuals derived from the selfing of F2 individuals designated +/+ (lanes 1–4); +/- (lanes 5–8); -/- (lanes 9–12) where + indicates AAT2 and - indicates aat2-1.

Genetic characterization of aat2 and aat3 mutants:
To determine whether the loss-of-activity gel phenotype observed in the aat2 mutants was controlled by a single nuclear gene, the F₂ generation (from crosses to wild-type Columbia plants) were analyzed for each of the three aat2 mutants. Three distinct gel phenotypes were observed in the F₂ generation for each of the aat2 mutants: wild-type (Figure 2A, lane 1), heterozygotes (+/-) with reduced levels of AAT2 (Figure 2A, lane 2), and homozygotes (-/-) with no detectable AAT2 activity (Figure 2A, lane 3). Each had normal levels of AAT3 activity. As the assay is quantitative, intermediate levels of AAT2 activity would be expected to occur in heterozygotes with structural gene mutations. To confirm that the F₂ individuals with the "intermediate" AAT2 gel phenotype were indeed heterozygotes, the putative heterozygotes were selfed and their F₃ progeny analyzed by gel assay. The F₃ generation from each of the putative heterozygotes showed the same segregation pattern as the F₂ generation, i.e., three distinct gel phenotypes could be discerned (Figure 2C, lanes 1–5 and 2E, lanes 5–8, show one example). By contrast, when F₂ individuals identified as +/+ or -/- were selfed, all F₃ individuals showed the gel phenotype of the parent (Figure 2E, lanes 1–4, and 9–12, respectively). By this criterion, all three aat2 mutants are judged to be semidominant because the phenotype (i.e., reduced AAT2 activity) observed in the heterozygotes is intermediate between the wild-type and homozygous mutant plants. Furthermore, genetic analysis of the segregating F₂ population is consistent with the mutant phenotypes being caused by a mutation in a single nuclear gene (Table 1). The aat2-1, aat2-2, and aat2-4 mutations are allelic as demonstrated by failure to complement in pairwise crosses, i.e., all individuals in the F₁ generation of the crosses between the aat2-1 mutant and the other aat2 mutants had the aat2 mutant gel phenotype (data not shown). For the aat3 mutants, it is also possible to distinguish heterozygous F₂ individuals for the mutation affecting AAT3 (Figure 2B, lane 2). When these putative heterozygotes are selfed, resulting F₃ individuals segregate for the AAT3 activity as +/+; +/-; and -/- (Figure 2D, lanes 1–5). Genetic analysis of the segregating F₂ populations is consistent with the aat3 mutant phenotypes being caused by mutations in a single nuclear gene (Table 1). The aat3-1 and aat3-2 mutations are allelic as demonstrated by failure to complement in pairwise crosses, i.e., all individuals in the F₁ generation of the crosses between the aat3-1 mutant and aat3-2 mutant had the aat3 mutant gel phenotype (data not shown).

Mapping of the five ASP genes and aat2 and aat3 mutant loci:
Genetic mapping was performed to determine whether the aat2 or aat3 mutants were linked to any of the five ASP genes (ASP1–ASP5) encoding distinct AspAT isoenzymes. For mapping purposes, several of the aat2 and aat3 mutants (Columbia ecotype) were outcrossed to the Landsberg (Ler) ecotype. The aat2 and aat3 mutant loci were then mapped relative to known CAPS or SSLP markers, using approximately 30 individuals, a number deemed sufficient to map each mutation relative to one of 19 markers on a specific arm of each chromosome (KONIECNZY and AUSUBEL 1993 ). Separately, RFLP or CAPS polymorphisms were identified for the ASP1–ASP5 genes and these ASP genes were mapped using Recombinant Inbred lines (for ASP1, 2, 3, and 5) or relative to known CAPS or SSLP markers (for ASP4). A summary of the RFLP or CAPS markers identified for each ASP gene is in MATERIALS AND METHODS. The relative map positions for the ASP1–ASP5 genes and aat2 and aat3 mutants are shown in Figure 3. This mapping data along with cosegregation analysis (see below) has enabled us to predict which ASP gene is likely to be affected in the aat2 or the aat3 mutants, as outlined below.

View larger version (19K): In this window In a new window Download PPT slide   Figure 3. —Map positions of the five AspAT genes (ASP1–ASP5) and the mutations present in plants deficient in cytosolic AAT2 (aat2) or chloroplastic AAT3 (aat3). ASP1–4 genes are described in SCHULTZ and CORUZZI 1995 . The ASP5 gene is described in WILKIE et al. 1995 , WILKIE et al. 1996 . The ASP gene locations and aat mutations are shown relative to a subset of markers on the recombinant inbred line map from October 1997 (Nottingham Arabidopsis Stock Centre, WWW server http://nasc.nott.ac.uk/). CAPS and SSLP markers used for mapping purposes are not italicized.

The aat3 mutation affecting the chloroplast AAT3 isoenzyme maps to the bottom of chromosome IV, near the PG11 CAPS marker (Figure 3). Arabidopsis contains two genes which could encode the major chloroplastic AAT3 isoenzyme: ASP3 (SCHULTZ and CORUZZI 1995 ) and ASP5 (WILKIE et al. 1995 , WILKIE et al. 1996 ). ASP3 was predicted to encode either a plastid or peroxisomal AspAT based on the presence of a putative transit peptide (SCHULTZ and CORUZZI 1995 ). ASP5 encodes an aspartate aminotransferase polypeptide which can be imported into chloroplasts, as judged by in vitro uptake experiments (WILKIE et al. 1995 ). Results presented here suggest that ASP5 most likely encodes the major chloroplast AAT3 isoenzyme in Arabidopsis, as it maps to chromosome IV at approximately 76 cM in the same location as the mutation in the aat3 plants deficient in chloroplast AAT3 (Figure 3). By contrast, the ASP3 gene maps to a separate chromosome (V) at approximately 25 cM.

The mutation affecting the major cytosolic AAT2 isoenzyme in the aat2 plants, mapped to the top of chromosome V (Figure 3). While there are two putative genes for cytosolic AAT2, ASP2 and ASP4, gene expression studies suggested that ASP2 is the likely candidate to encode the major cytosolic AAT2 isoenzyme. ASP2 mRNA accumulates to high levels especially in roots, while ASP4 mRNA is expressed at extremely low levels in all tissues examined (SCHULTZ and CORUZZI 1995 ). In support of this, the aat2-4 mutation and the ASP2 gene each map to the same region of chromosome V. By contrast, the ASP4 gene maps to a different chromosome (chromosome I). Independent genetic evidence also suggests that the ASP2 gene and the aat2-4 mutation are linked. All 33 homozygous aat2-4 mutants identified from a segregating F₂ population from the cross, aat2-4/aat2-4 (Col) x AAT2/AAT2 (Ler), showed the Columbia-specific RFLP for ASP2 (data not shown).

aat2-2 mutants deficient in cytosolic AAT2 display reduced growth rate and aspartate deficiency:
Phenotypic analysis of the aat mutants deficient in either cytosolic AAT2 (aat2-1, aat2-2, aat2-4) or chloroplastic AAT3 (aat3-1 and aat3-2) provides a means to analyze the in vivo role of each of the two major AspAT isoenzymes in plant nitrogen metabolism. As all the aat2 and aat3 mutants were isolated and propagated on media or soil supplemented with 0.05% aspartate, we first determined whether any of these mutants were auxotrophic for aspartate. In all cases, the seed from the aat2 and aat3 mutants germinated and the developing plants set seed in the absence of any amino acid supplement under normal growth conditions (data not shown). These results suggest that none of the aat2 or aat3 mutants are auxotrophic. To determine whether any of the aat2 or aat3 mutants exhibit more subtle growth impairments, growth rate of the mutant plants was compared to wild type. For this, aat2 or aat3 mutants were sown side-by-side with wild type on tissue culture plates containing MS media supplemented with 3% sucrose. Plates were incubated vertically, and root length was measured as an indicator of growth rate. Neither of the aat3 mutants (aat3-1 or aat3-2) displayed impaired growth. Of the three aat2 mutants, only the aat2-2 mutant showed a reduction in root growth (20–50%), compared to wild-type Columbia controls (Figure 4). To minimize the possibility that the reduced rate of root growth in aat2-2 was due to a background mutation, three independent homozygous mutant lines and three homozygous wild-type lines from the F₂ generation of the cross aat2-2/aat2-2 x AAT2/AAT2 (Col) were tested in a repeat experiment. All three of the homozygous mutant lines (aat2-2/aat2-2) showed the reduced growth rate phenotype, whereas all three of the wild-type (AAT2/AAT2) lines from the segregating population exhibited wild-type root growth rates (data not shown).

View larger version (153K): In this window In a new window Download PPT slide   Figure 4. —aat2-2 mutants exhibit reduced growth rates. The growth of aat2-2 mutant plants and wild-type (Col) plants were monitored on MS medium containing 3% sucrose.

The altered growth of the aat2-2 mutant suggested that the AAT2 isoenzyme might play a major role in assimilation of primary nitrogen into aspartate, as aspartate serves to transport assimilated nitrogen in many plant species including Arabidopsis (URQUHART and JOY 1981 ; RICHARDSON and BAKER 1982 ; HAYASHI and CHINO 1986 ; SCHULTZ 1994 ; LAM et al. 1995 ). HPLC analysis showed that the aat2-2 mutant indeed contains an 80% reduction in levels of aspartate transported in phloem exudates of light-grown aat2-2 plants (Figure 5). Levels of free aspartate were also specifically reduced in whole leaf extracts of light-grown plants (SCHULTZ 1994 ). In addition, the aat2-2 mutant showed a specific and significant reduction (50%) in the levels of free asparagine measured in phloem exudates of dark-adapted plants (Figure 5). These results suggest that the pool of aspartate synthesized by cytosolic AAT2 in the light is converted to asparagine when plants are dark adapted. As levels of aspartate are unaffected in dark-adapted aat2-2 mutants compared to wild type, this suggests another AAT isoenzyme controls aspartate levels in dark-adapted plants. The other two aat2 mutants (aat2-1 and aat2-3) and the two aat3 mutants (aat3-1 and aat3-2) each showed normal amino acid profiles.

View larger version (16K): In this window In a new window Download PPT slide   Figure 5. —aat2-2 mutants have specific reductions in levels of aspartate in light-grown plants and asparagine in dark-adapted plants. The relative proportions of aspartate and asparagine in the phloem exudates from wild-type Columbia (Col) and aat2-2 mutant plants grown in light (unshaded boxes) or dark adapted (shaded boxes). Each sample is the average of a single leaf from three representative plants. Plants were grown in soil in a normal day/night cycle (16 hr light/8 hr dark) for 3 wk and either light adapted () or dark adapted () for 24 hr. Error bars represent the standard error of the mean.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In plants, the amino acid aspartate is used to transport assimilated nitrogen from sources to sinks. As such, aspartate is one of the four most abundant free amino acids in leaves of many plant species including Arabidopsis and is also a major amino acid transported in the phloem (SCHULTZ 1994 ; LAM et al. 1995 ). Thus, understanding which isoenzymes of aspartate aminotransferase control the synthesis of aspartate in higher plants has significance for plant nitrogen use. As there are five ASP genes encoding isoenzymes of AspAT localized to the cytosol, chloroplast, mitochondria, or peroxisomes, we used plant mutants defective in specific AspAT isoenzymes to determine the in vivo function of each in aspartate biosynthesis. Historically, the use of mutants has been a powerful tool in the study of amino acid biosynthesis in microorganisms. However, the isolation of amino acid biosynthesis mutants in plants has been hampered by issues such as gene redundancy and problems with uptake of supplemented amino acids. There are only a few examples of whole plant mutants in amino acid biosynthesis enzymes. For example, Arabidopsis tryptophan biosynthesis mutants were isolated using a positive selection for resistance to 5-methylanthranilate (5-MA) (LAST and FINK 1988 ). In this case, gene redundancy does occur, but a mutation in the more highly expressed gene for tryptophan synthase ß (TBS1) can lead to auxotrophy (LAST et al. 1991 ). Whole plant mutants have also been isolated with specific defects in isoenzymes of glutamine synthetase (GS2) (WALLSGROVE et al. 1987 ) or glutamate synthase (Fd-GOGAT) (SOMERVILLE and OGREN 1980 ). These mutants were isolated based on their conditional lethal phenotype: death under photorespiratory growth conditions and growth in elevated CO₂. Mutants deficient in chloroplastic isoenzymes of GS2 or Fd-GOGAT demonstrated that photorespiratory ammonia released in mitochondria is reassimilated by isoenzymes localized in the chloroplast. As the factors controlling the flow of metabolites between subcellular compartments are as yet unknown, these case studies highlight the power of using genetics to define the metabolic flux that occurs between organelles in planta.

In the above examples, whole plant mutants were selected or screened for on the basis of a phenotype: resistance to 5-methylanthranilate (5-MA) or photorespiratory defects. In the case of aspartate aminotransferase (AspAT) where five genes encode isoenzymes localized to four distinct subcellular compartments, it was impossible to predict whether a mutation in any one ASP gene would lead to an associated aspartate-related phenotype. This is especially so given that other aminotransferases such as tyrosine aminotransferase might functionally mask a genetic defect in an aspartate aminotransferase gene (WIGHTMAN and FOREST 1978 ). Since it was impossible to predict a phenotype (if any) resulting from a mutation in any one of the five ASP genes, we screened for Arabidopsis mutants deficient in specific AspAT isoenzymes using a brute-force screen for loss of isoenzyme activity. In this screen, aspartate was added to the growth media to allow the isolation of putative auxotrophs. Once the aat mutants were identified on the supplemented media, we then performed phenotypic analysis to assay for any aspartate related defects.

In a screen of 8000 M₂ seedlings, mutants defective in either of the two major AspAT isoenzymes were isolated: mutants lacking either cytosolic AAT2 (aat2, three mutant alleles) or chloroplastic AAT3 (aat3, two mutant alleles). The mutations affecting cytosolic AAT2, map to the same location and cosegregate with the ASP2 gene, one of two genes for cytosolic AspAT. The mutations affecting chloroplastic AAT3 map to the same local region as the ASP5 gene, a nuclear gene coding for chloroplastic AspAT (WILKIE et al. 1995 , WILKIE et al. 1996 ). This mapping and cosegregation data support the notion that the mutations affecting cytosolic AAT2 or chloroplastic AAT3 are likely to be due to lesions in the structural genes for AspAT, ASP2 and ASP5, respectively. The dosage effect observed in both the aat2 and aat3 heterozygotes (intermediate levels of isoenzyme activity) would be expected from mutations in structural genes. Furthermore, preliminary analysis indicates that all aat2 mutants have normal levels of ASP2 mRNA, supporting the notion that the mutations affecting the aat2 isoenzyme are structural rather than regulatory gene mutations (data not shown). Sequencing of specific ASP genes in each mutant will verify whether the loss of AAT activity is due to a structural gene mutation.

Phenotypic analysis revealed that the majority of the aat2 and aat3 mutants (with the notable exception of aat2-2) do not exhibit growth impairments or aspartate deficiencies, suggesting a significant degree of functional redundancy among the AspAT isoenzymes in Arabidopsis. An alternate explanation is that most of the mutants are leaky and contain sufficient residual AspAT activity (or other aminotransferase activities) to permit normal growth. It is, however, noteworthy that the aat3 and aat2 mutants contain less than 3–6% wild-type activity, respectively, as determined by native gel assay. Another possibility is that there are other phenotypes that have not yet been uncovered. The finding that all three aat2 mutants have low levels of enzyme activity detected in vitro (less than 6% wild type) but only one (aat2-2) shows an aspartate deficiency and growth phenotype is reminiscent of the Arabidopsis trp1 mutants. trp1 mutants have defects in the enzyme phosphoribosylanthranilate transferase (PAT). Nine allelic trp1 mutants show undetectable levels of activity in vitro (less than 1%); however, only four of these trp1 mutants require tryptophan for growth (auxotrophs) while the others do not (prototrophs) (ROSE et al. 1997 ).

The aat2-2 mutants deficient in cytosolic AAT2, display defects in growth and a specific and dramatic reduction in the levels of free and transported aspartate. It is notable that the defect in aspartate synthesis in the aat2-2 mutant is conditional on light. That is, light-grown aat2-2 plants exhibit an 80% decrease in levels of transported aspartate. By contrast, levels of free aspartate are unaffected in dark-adapted plants. These results suggest that the cytosolic AAT2 isoenzyme controls the bulk of aspartate synthesized in the light, and suggests that another AspAT isoenzyme controls aspartate synthesized in the dark. Moreover, the aat2-2 mutant plants also show a dramatic decrease in levels of free and transported asparagine, specifically in dark-adapted plants. This finding indicates that aspartate synthesized in the light by the cytosolic AAT2 isoenzyme supplies the pool of aspartate used for asparagine synthesis in the dark. This metabolic conversion of aspartate to asparagine appears to reflect carbon:nitrogen economy in plants. In the dark, when carbon skeletons are limiting, asparagine (2N:4C) serves as a more carbon-efficient nitrogen transport compound compared to aspartate (1N:4C), glutamate (1N:5C), or glutamine (2N:5C) (see Figure 6). These findings provide in vivo support for a metabolic control model proposed by LAM et al. 1995 in which aspartate synthesized in the light was predicted to be the precursor to asparagine synthesized in the dark (LAM et al. 1994 ; LAM et al. 1995 ). While this model was initially developed based on the preferential synthesis of asparagine in the dark and by the differential regulation of nitrogen assimilation genes by light and metabolites, the aat2-2 mutants provide experimental evidence suggesting this mechanism is operating in vivo. Moreover, the aat2 mutants identify cytosolic AAT2, as the specific isoenzyme controlling the synthesis of this pool of aspartate. The finding that cytosolic AAT2 controls the bulk of aspartate synthesized in the light is somewhat unexpected. As nitrogen is assimilated into glutamate in plastids (by GOGAT), based on subcellular compartmentation, chloroplastic AAT3 would be a likely candidate to control aspartate synthesis in the light. The lack of a detectable aspartate-related phenotype associated with either of the two mutants defective in chloroplastic AAT3 cannot exclude a major role for this isoenzyme in aspartate synthesis, as neither of the two aat3 mutants may contain a null allele. Nonetheless, the aat2-2 mutants demonstrate that cytosolic AAT2, not chloroplastic AAT3, controls the bulk of aspartate synthesized in the light for nitrogen transport.

View larger version (78K): In this window In a new window Download PPT slide   Figure 6. —Cytosolic AAT2 controls the synthesis of aspartate in the light, which is converted to asparagine in the dark. A model is depicted for the metabolic flow of nitrogen assimilation into the nitrogen-transport amino acids glutamate, glutamine, aspartate, and asparagine in the light and dark. In the light, inorganic nitrogen is assimilated initially into glutamate and glutamine by the combined actions of the plastid enzymes: chloroplastic glutamine synthetase (GS2, encoded by GLN2), and ferredoxin-dependent glutamate synthase (Fd-GOGAT, encoded by GLU1; OLIVEIRA et al. 1997 ; COSCHIGANO et al. 1998 ). The conversion of glutamate into aspartate in the light is controlled by cytosolic AAT2. In the dark, this pool of aspartate is converted into asparagine by asparagine synthetase (ASN1) (LAM et al. 1994 , LAM et al. 1995 ).

The preliminary analysis of aat2 mutants described herein shows that cytosolic AAT2 plays an important role in nitrogen assimilation into aspartate in light-grown plants. In addition, the aat2-2 mutant also identifies two distinct features about the genes for cytosolic AAT2. First, while there are two genes for cytosolic AspAT in Arabidopsis (ASP2 and ASP4), a mutation linked to one gene leads to a phenotype. This is reminiscent of the case for duplicated genes for tryptophan synthase (LAST et al. 1991 ), as ASP2 is the major expressed gene for cytosolic AspAT, while ASP4 is expressed at extremely low levels (based on steady state levels of mRNA; SCHULTZ and CORUZZI 1995 ). Thus, the ASP4 gene may function at low constitutive levels to provide aspartate for protein synthesis, while ASP2 serves to synthesize aspartate used as a nitrogen transport amino acid in the phloem. The aat2 mutants also indicate that the major cytosolic AAT2 isoenzyme is functionally distinct from the AspAT isoenzymes which occur in the mitochondria, chloroplasts, and peroxisomes. While these other subcellular isoenzymes may be involved in shuttling reducing equivalents between subcellular compartments, it appears that the cytosolic AAT2 isoenzyme controls the bulk of the assimilation of nitrogen into aspartate used to transport nitrogen within the plant. Thus, these genetic studies have pinpointed the AAT2 isoenzyme as the isoenzyme controlling the synthesis of transported aspartate. As aspartate is a precursor to essential amino acids including lysine, these basic research studies using a genetic approach in Arabidopsis to identify isoenzymes controlling metabolic flux into transported aspartate may have significance for modifying levels of aspartate-derived amino acids in transgenic crop plants.

	FOOTNOTES

¹ Present address: Co-operative Research Centre for Industrial Plant Biopolymers, School of Botany, The University of Melbourne, Parkville, 3052, Australia.

	ACKNOWLEDGMENTS

This work was supported by National Science Foundation (NSF) grant MCB-9304913. An NSF RAHMMS supplement was awarded to M. HSU who was also awarded a Westinghouse semi-finalist prize for her part of this project. An NSF REU was awarded to ARTEM VAYNMAN who performed some of the mutant screens. Thanks to KAREN COSCHIGANO for helpful discussions during the course of this project and to ALEXANDRA CLARK for performing some of the RI mapping. We also thank IGOR OLIVEIRA for assistance in preparation of some figures and LIVIA WEI for help with the initial screening of the mutants.

Manuscript received January 9, 1998; Accepted for publication March 4, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BELL, C. J. and J. R. ECKER, 1994  Assignment of 30 microsatellite loci to the linkage map of Arabidopsis.. Genomics 18:137-144.

CARVALHO, H., S. PEREIRA, C. SUNKEL, and R. SALEMA, 1992  Detection of cytosolic glutamine synthetase in leaves of N. tabacum L. by immunocytochemical methods. Plant Physiol. 100:1591-1594[Abstract/Free Full Text].

COSCHIGANO, K. T., R. MELO-OLIVEIRA, J. LIM, and G. M. CORUZZI, 1998  Arabidopsis gls mutants and distinct Fd-GOGAT genes: implications for photorespiration and primary nitrogen assimilation. Plant Cell 10:1-12[Free Full Text].

EDWARDS, J. W., E. L. WALKER, and G. M. CORUZZI, 1990  Cell-specific expression in transgenic plants reveals nonoverlapping roles for chloroplast and cytosolic glutamine synthetase. Proc. Natl. Acad. Sci. USA. 87:3459-3463[Abstract/Free Full Text].

FORDE, B. G., H. M. DAY, J. F. TURTON, W. J. SHEN, and J. V. CULLIMORE et al., 1989  Two glutamine synthetase genes from Phaseolus vulgaris L. display contrasting developmental and spatial patterns of expression in transgenic Lotus corniculatus plants. Plant Cell 1:391-401[Abstract/Free Full Text].

GIVAN, C. V., 1990 Aminotransferases in higher plants, pp. 329–357 in The Biochemistry of Plants, Vol. 5, edited by P. K. STUMPF and E. E. CONN. Academic Press, New York.

HAYASHI, H. and M. CHINO, 1986  Collection of pure phloem sap from wheat and its chemical composition. Plant Cell Physiol. 27:1387-1393[Abstract/Free Full Text].

IRELAND, R. J., and K. W. JOY, 1985 Plant transaminases, pp. 376–384 in Transaminases, edited by P. CHRISTEN, and D. E. METZLER. John Wiley and Son, New York.

KONIECNZY, A. and F. AUSUBEL, 1993  A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers. Plant J. 4:403-410[Medline].

LAM, H.-M., S. S.-Y. PENG, and G. M. CORUZZI, 1994  Metabolic regulation of the gene encoding glutamine-dependent asparagine synthetase in Arabidopsis thaliana.. Plant Physiol. 106:1347-1357[Abstract].

LAM, H.-M., K. COSCHIGANO, C. SCHULTZ, R. MELO-OLIVEIRA, and G. TJADEN et al., 1995  Use of Arabidopsis mutants and genes to study amide amino acid biosynthesis. Plant Cell 7:887-898[Medline].

LAST, R. L. and G. R. FINK, 1988  Tryptophan requiring mutants of the plant Arabidopsis thaliana.. Science 240:305-310[Abstract/Free Full Text].

LAST, R. L., P. H. BISSINGER, D. J. MAHONEY, I. E. R. RADWANSKI, and G. R. FINK, 1991  Tryptophan mutants in Arabidopsis: the consequences of duplicated tryptophan ß synthase genes. Plant Cell 3:345-358[Abstract/Free Full Text].

LISTER, C. and C. DEAN, 1993  Recombinant inbred lines for mapping RFLP and phenotypic markers in Arabidopsis thaliana.. Plant J. 4:745-750.

LIU, K. D. F. and A. H. C. HUANG, 1977  Subcellular localization and developmental changes of aspartate--ketoglutarate transaminase isozymes in the cotyledons of cucumber seedlings. Plant Physiol. 59:777-782[Abstract/Free Full Text].

MEINKE, D. and M. KOORNNEEF, 1997  Community standards for Arabidopsis genetics. Plant J. 12:247-253.

OLIVEIRA, I. C., H.-L. LAM, K. COSCHIGANO, R. MELO-OLIVEIRA, and G. CORUZZI, 1997  Molecular-genetic dissection of ammonium assimilation in Arabidopsis thaliana.. Plant Physiol. Biochem. 35:185-198.

PETERMAN, T. K. and H. M. GOODMAN, 1991  The glutamine synthetase gene family of Arabidopsis thaliana: light-regulation and differential expression in leaves, roots and seeds. Mol. Gen. Genet. 230:145-154[Medline].

REDEI, G. P., and C. KONCZ, 1992 Classical mutagenesis, pp. 16–82 in Methods in Arabidopsis Research, edited by C. KONCZ, N.-H. CHUA and J. SCHELL. World Scientific, Singapore.

RICHARDSON, P. T. and D. A. BAKER, 1982  Cellular localization of nodule-enhanced aspartate aminotransferase. J. Exp. Bot. 33:1239-1247[Abstract/Free Full Text].

ROSE, A. B., J. LI, and R. L. LAST, 1997  An allelic series of blue fluorescent trp1 mutants of Arabidopsis thaliana.. Genetics 145:197-205[Abstract].

SCHULTZ, C. J., 1994 A molecular and genetic dissection of the aspartate aminotransferase isoenzymes of Arabidopsis thaliana. Ph.D. Thesis, New York University.

SCHULTZ, C. J. and G. M. CORUZZI, 1995  The aspartate aminotransferase gene family of Arabidopsis encodes isoenzymes localized to three distinct subcellular compartments. Plant J. 7:61-75[Medline].

SOMERVILLE, C. and W. OGREN, 1980  Inhibition of photosynthesis in Arabidopsis mutants lacking leaf glutamate synthase activity. Nature 286:257-259.

URQUHART, A. A. and K. W. JOY, 1981  Use of phloem exudate technique in the study of amino acid transport in pea plants. Plant Physiol. 68:750-754[Abstract/Free Full Text].

WALLSGROVE, R. M., J. C. TURNER, N. P. HALL, A. C. KENDALL, and W. S. BRIGHT, 1987  Barley mutants lacking chloroplast glutamine synthetase—biochemical and genetic analysis. Plant Physiol. 83:155-158[Abstract/Free Full Text].

WEEDEN, N. F. and G. A. MARX, 1987  Further genetic analysis and linkage relationships of isozyme loci in pea. J. Hered. 78:153-159[Abstract/Free Full Text].

WENDEL, J. F., and N. F. WEEDEN, 1989 Genetics of plant isozymes, pp. 46–72 in Isozymes in Plant Biology, edited by D. E. SOLTIS and P. E. SOLTIS. Dioscorides Press, Portland, OR.

WILKIE, S. E., J. M. ROPER, A. G. SMITH, and M. J. WARREN, 1995  Isolation, characterization and expression of a cDNA clone encoding aspartate aminotransferase from Arabidopsis thaliana.. Plant Mol. Biol. 27:1227-1233[Medline].

WILKIE, S. E., R. LAMBERT, and M. J. WARREN, 1996  Chloroplastic aspartate aminotransferase from Arabidopsis thaliana—an examination of the relationship between the structure of the gene and the spatial structure of the protein. Biochem. J. 319:969-976.

WIGHTMAN, F. and J. C. FOREST, 1978  Properties of plant aminotransferases. Phytochemistry 17:1455-1471.