Results

Plants Have Genes Specifying PabA-PabB Hybrid Proteins. blast searches of GenBank and TIGR databases with the protein sequences of E. coli PabA and PabB detected a single Arabidopsis gene (At 2 g 28880) and a cognate cDNA (RAFL 09-32-D04) encoding a 103-kDa polypeptide with PabA- and PabB-like domains. Truncated tomato expressed sequence tags specifying a similar protein were also found; 5′ RACE was used to obtain the missing sequence, and complete tomato cDNAs encoding a 102-kDa protein were then isolated by RT-PCR. The deduced Arabidopsis protein is shown in Fig. 2 and is aligned with its tomato counterpart in
Fig. 7, which is published as supporting information on the PNAS web site. Both plant proteins comprise tandem domains that share 33-37% identity with E. coli PabA and PabB, and 26% identity with yeast PABA synthase (Figs. 2 A and 7). The plant
proteins share 61% overall identity with each other. As in yeast and other PABA synthases, the PabA and PabB domains of the plant proteins are separated by a long linker (≈90 residues) whose sequence is not conserved but is rich in basic amino acids (Figs. ​(Figs.2B2B and 7). In contrast to yeast PABA synthase, the plant proteins have
a ≈45-residue insertion near the end of the PabA domain. Like the interdomain linker, this extra sequence is not conserved between Arabidopsis and tomato but is basic in character. The plant proteins also have nonconserved N-terminal extensions of ≈85 residues that display the features of chloroplast targeting peptides (Figs. ​(Figs.22 an) fig. 2.

Primary structure of the deduced Arabidopsis PabA-PabB polypeptide. (A) Scheme comparing the organization of the Arabidopsis polypeptide with that of eukaryotic PABA synthases and with E. coli PabA and PabB. Interdomain linker, red; insertion in PabA (more ...)







Plant PabA-PabB cDNAs Complement E. coli and Yeast PABA Synthesis Mutants. To determine whether the Arabidopsis and tomato cDNAs encode functional enzymes, the sequences predicted to specify the mature proteins (Figs. ​(Figs.2B2B and 7) were cloned into E. coli expression vector pLOI707HE and yeast
expression vector pVT103-U. pLOI707HE allows tightly IPTG-controlled gene expression in diverse host strains (18). The resulting constructs were transformed into an E. coli pabA pabB double mutant and a yeast PABA synthase deletant, both of which are PABA auxotrophs (Fig. 3A). The Arabidopsis and tomato constructs
yielded PABA-independent transformants of the E. coli and yeast mutants with high frequency, and their growth was similar to that of wild-type strains (Fig. 3A). No
complementation was seen with vectors alone (Fig. 3A). In E. coli harboring plant
cDNA constructs, complementation depended strictly on addition of the IPTG inducer, indicating that the plant cDNAs were responsible for the effect (Fig. 3A).
Retransformation of the yeast mutant with rescued plasmid containing the Arabidopsis cDNA restored PABA prototrophy, further confirming that complementation was due to the plant cDNA (data not shown). Taken together, these data indicate that plant PabA-PabB proteins have PabA plus PabB activity (i.e., that they are functional ADCSs) but say nothing about their PabC (ADC lyase) activity.

Fig. 3.
Plant PabA-PabB proteins encode functional enzymes. (A) Complementation of a yeast PABA synthase mutant (Upper) and an E. coli pabA pabB double mutant (Lower). Similar numbers of cells of wild-type (WT) yeast (strain 971/6c) or E. coli (strain K12) and (more ...)


Arabidopsis PabA-PabB Has ADC Synthase but Not Lyase Activity. To investigate ADCS and lyase activities directly, the mature Arabidopsis PabA-PabB protein was overexpressed in E. coli by using the pET-28a plasmid and the host strain BL21-CodonPlus (DE3)-RIL (which is wild type with respect to pabA, pabB, and pabC). ADCS activity was first assayed in desalted extracts, by using glutamine as amino donor and adding excess E. coli PabC to convert ADC to PABA, which was quantified by HPLC. In this coupled assay, vector-alone cell extracts produced traces of PABA (Fig. 3B), attributable to the endogenous PabA and PabB activities of the
host. Extracts of cells overexpressing the Arabidopsis PabA-PabB protein gave 20-fold more PABA (Fig. 3B), providing initial biochemical evidence for ADCS activity.
To corroborate this result and to test for ADC lyase activity, we used size-exclusion chromatography to separate the recombinant Arabidopsis PabA-PabB protein from the E. coli host's PabC enzyme, which is smaller (≈50 kDa; ref. 19). By using glutamine as amino donor, the column fractions were assayed for PABA formation, plus or minus addition of E. coli PabC. When PabC was added, there was a large peak of PABA-forming activity at ≈110 kDa (fractions 19-20), which is close to the predicted mass of the monomeric recombinant enzyme (93 kDa) (Fig. 4A). This
peak disappeared when PabC was omitted (Fig. 4A) and was not given by extracts of
cells transformed with vector alone (Fig. 4B). The minor peak of PABA formation in
fractions 23-24 is attributable to endogenous activity from the host cells (Fig. 4 A and B). These data show that the Arabidopsis PabA-PabB protein has
ADCS activity but lacks ADC lyase activity. The Arabidopsis and tomato proteins may thus be termed ADCSs (AtADCS and LeADCS, respectively). The AtADCS peak of Fig. 4A was unstable, losing all activity within 12 h at 4°C.

Fig. 4.
The Arabidopsis PabA-PabB protein has ADCS but not ADC lyase activity. Extracts of E. coli cells overexpressing the Arabidopsis PabA-PabB protein (A) or harboring the vector alone (B) were fractionated by size exclusion chromatography. Fractions were (more ...)





Amino Donors and Inhibitors of Arabidopsis ADC Synthase. When testing amino donors, we compared the Arabidopsis enzyme with the E. coli enzyme, using assay conditions developed for the latter (19). Like PabA, AtADCS did not use asparagine (Table 1). AtADCS was able to use NH3, like PabB, but apparently more efficiently because the NH3-dependent rate was 50% of the glutamine-dependent rate for AtADCS but only 9% for PabB (Table 1). AtADCS was not inhibited by physiological concentrations of PABA (0.5 μM), PABA glucose ester (6 μM), or folates (10 μM tetrahydrofolate, 5-methyltetrahydrofolate, or 5-formyltetrahydrofolate pentaglutamate), nor was it sensitive to the PABA analogs PABA ethyl ester, p-acetamidobenzoate, p-hydroxybenzoate, or anthranilate (50 μM) (data not shown).


Table 1.
Activities of Arabidopsis and E. coli ADC synthases with various amino donors

Subcellular Localization of Arabidopsis ADC Synthase. To analyze the subcellular localization of AtADCS in vivo, its N-terminal region (85 residues) was fused to the GFP marker protein. Expression of this fusion protein in Arabidopsis protoplasts resulted in a punctate pattern of green fluorescence that colocalized with the red autofluorescence of chlorophyll (Fig. 5 E and F). In contrast, control
protoplasts expressing GFP alone showed green fluorescence throughout the cytoplasm and nucleus (Fig. 5C). These data demonstrate that AtADCS contains a
functional chloroplast targeting peptide, suggesting that plant ADCS is a plastidial enzyme.

Fig. 5.
Expression in Arabidopsis protoplasts of GFP fused to the N-terminal region (residues 3-87) of the Arabidopsis PabA-PabB protein and of GFP alone. Images are optical photomicrographs (A and D), chlorophyll fluorescence (B and E, red pseudocolor), and (more ...)

ADC Synthase mRNA Expression During Tomato Fruit Development. Because nothing is known about PABA synthesis in plants, we measured LeADCS transcript expression and total PABA pools in leaves and in fruits harvested at four stages. LeADCS mRNA levels in unripe fruits were initially ≈10% of that in leaves but dropped dramatically to beneath the detection limit after the breaker stage (Fig. 6A). The
detection of the Nr transcript, which is induced in ripening fruit (25), showed that this fall was not due to general mRNA degradation as ripening advanced (Fig. 6A Inset).
Concentrations of PABA in the fruit increased progressively as the fruit ripened from the mature green to the red and red-ripe stages (Fig. 6B). As previously observed
(24), the majority of PABA (80-90%) was present as its glucose ester (data not shown).

Fig. 6.
Levels of LeADCS mRNA, PABA, and folate in tomato fruits and in leaves. (A) LeADCS mRNA was quantified by real-time RT-PCR on 250 ng of total RNA from pericarp tissue of fruits at mature green (MG), breaker (B), ripe (R), and red-ripe (RR) stages, or (more ...)