Microbial Origin of Plant-Type 2-Keto-3-Deoxy-d-arabino-Heptulosonate 7-Phosphate Synthases, Exemplified by the Chorismate- and Tryptophan-Regulated Enzyme from Xanthomonas campestris

Bacterial strains, plasmids, media, and growth conditions.

The Xanthomonas campestris and E. coli strains and plasmids used in this work are listed in Table 1. Growth media for X. campestris and E. coli included Luria broth (LB) as a complete medium. ARO minimal medium is a modification of the medium reported by Ray et al. (24). It had the following composition (in grams per liter): glucose (1), K₂HPO₄ (7), KH₂PO₄ (2), (NH₄)₂SO₄ (0.5), ferric ammonium citrate (0.32), NaCl (0.5), and Casamino Acids (5). After autoclaving, the following compounds (grams per liter) were added: p-aminobenzoic acid (0.01), p-hydroxybenzoic acid (0.01), 2,3-dihydroxybenzoic acid (0.01), thiamine-HCl (0.0005), and 1 ml of 1 M MgSO₄. For induction of the tac promoter, IPTG (isopropyl-β-d-thiogalactopyranoside) was added to give a final concentration of 0.2 mM. Ampicillin was used when required at a final concentration of 100 μg/ml.

Materials.

Enzymes for molecular genetic applications were purchased from New England BioLabs or Boehringer Mannheim and were used according to the specifications of the manufacturer. Chorismate purified from the accumulation medium of the triple auxotroph Klebsiella pneumoniae 62-1 (ATCC 25306) was prepared as the free acid (12) and was 97% pure. ZnSO₄, MgCl₂, NiCl₂, and MnCl₂ (puratronic grade) were obtained from Johnson Matthey (Ward Hill, Mass.); other metal salts (reagent grade) were from Sigma (St. Louis, Mo.). EDTA (analytical reagent grade) was purchased as the disodium salt from Mallinckrodt (Paris, Ky.). High-grade Spectra/Por dialysis tubing (VWR) was boiled in 2 mM EDTA and exhaustively washed before use. Chelex-treated water was used for preparation of fresh metal solutions. FeSO₄ solutions were always prepared immediately before assays were performed by dissolving it in 0.01 N trace metal grade HCl (Fisher Scientific). Enzyme grade ammonium sulfate was from Sigma.

General DNA techniques and sequencing.

Cloning experiments were carried out according to the methods described by Sambrook et al. (27). The DNA sequence of aroA_II and flanking 5′ and 3′ regions was determined from plasmid pTacIQ and selected subclones and by primer walking by the dideoxy chain termination procedure (28) using a Taq FS dye terminator fluorescence-based cycle sequencing kit (Applied Biosystems) with a Perkin-Elmer/Applied Biosystems Model 377-18E1 sequencer.

Preparation and screening of a genomic DNA library.

Total genomic DNA from X. campestris was prepared following the method described by Silhavy et al. (31) and was partially digested with Sau3AI. Genomic DNA fragments from about 1 to 7 kb in length were isolated from an agarose gel and ligated to BamHI-digested pTacIQ. We had observed that direct transformation of an X. campestris genomic library into an McrA and McrB strain resulted in a lower number of transformants than in a strain without an Mcr system. For this reason, the ligation mixture was used to transform the mcrA mcrCB E. coli strain XL1 Blue MR. Plasmid DNA from the XL1 Blue MR library was purified and used to transform strain AB3248. Transformants were selected on ARO medium plates containing 100 μg of ampicillin per ml and 0.2 mM IPTG.

Strains and growth of cultures.

E. coli AB3248, a triple mutant deficient for all three DAHP synthase paralogs, was grown on LB or on minimal medium plus aromatic amino acids required for growth. E. coli DH5α and E. coli BL21(DE3) were grown on LB plus kanamycin when transformed with a plasmid. X. campestris ATCC 33436 (wild type) was grown in either LB or minimal medium supplemented with trace minerals and aromatic amino acids where appropriate.

Clones and plasmids.

Plasmid pTacIQ2.3 contains a 6.5-kb DNA segment from X. campestris. Plasmid pTacIQ2.3B contains a 2.7-kb fragment of X. campestris DNA from plasmid pTacIQ2.3, which carries the gene for DAHP synthase and complementing activities. Plasmid pTacIQ2.3B was used to prepare the pET-24b⁺ plasmid derivatives pET-M1 and pET-M2 carrying the M1 and M2 clones of DAHP synthase, respectively.

Preparation of clones M1 and M2.

Oligonucleotides containing an NdeI restriction site were made from the two possible 5′ ATG starts of the DAHP synthase nucleotide sequence, DAHPAI (5′-GATAAACATATGGGTCCGGCTGCGGCCGGCGTC-3′) and DAHPBI (5′-GATAAACATATGTCCGCTTCTTCCCCCTCGAC-3′). An oligonucleotide denoted DAHPSAII with an EcoRI restriction site and corresponding in reverse to the 3′ end of the sequence was also made (5′-GATAAGAATTCCTGCGCAGTCGTTTTGATC-3′). PCRs which included one each of the 5′ oligonucleotides and the 3′ oligonucleotides and the 2.8-kb fragment were performed. The PCR products were gel purified and subjected to digestion with NdeI and XhoI. The same restriction endonuclease digestion was performed on the pET-24b⁺ plasmid, which was then purified by gel electrophoresis. The PCR fragments were ligated with the plasmid, transformed into DH5α cells, and spread on LB-kanamycin plates. Several colonies were chosen for analysis. DNA from plasmids containing M1 (pET-M1) and M2 (pET-M2) start sites was purified, subjected to nucleotide sequencing to verify the correct sequence for each, and transformed into BL21(DE3) competent cells to obtain strains designated M1 and M2. A colony from each strain was inoculated into 10 ml of LB-kanamycin for overnight growth at 30°C. These cultures were used to inoculate 10 ml of LB-kanamycin the following morning. IPTG was added when cells reached a density of about 0.6 at A₆₀₀ and were harvested about 2.5 h later. Pellets were used immediately to prepare extracts.

Extract preparation.

Cell pellets harvested from M1, M2, X. campestris, or E. coli AB3248 were suspended into 100 mM potassium phosphate buffer at pH 7.5 containing 1.0 mM DTT, 1.0 mM benzamidine, and 0.1 mM phenylmethylsulfonyl fluoride. Cell suspensions were passed at 18,000 1b/in² through a French pressure cell, and the debris was pelleted by centrifugation at 20,000 × g for 30 min. Crude extracts to be used for enzyme assays were desalted on 10DG columns (Bio-Rad).

Enzyme assays.

DAHP synthase activities were assayed by use of the thiobarbituric acid assay as described before (15). Unless indicated otherwise, assays were carried out at 37°C in 50 mM N-(2-hydroxyethyl)piperazine-N′-(3-propanesulfonic acid) (EPPS) buffer at pH 8.5.

Enzyme partial purification.

For purification, a 0 to 25% ammonium sulfate fraction precipitate was centrifuged in a Sorval centrifuge for 20 min at 4°C. The supernatant fluid was further fractionated to 45% ammonium sulfate and centrifuged. The remaining supernatant was brought to 70% saturation with ammonium sulfate and centrifuged. All pellets were suspended in 50 mM EPPS buffer at pH 8.5, dialyzed for 6 h at 4°C with three buffer changes, and assayed for enzyme activities. Most of the activity was found in the 25 to 45% precipitate fraction, but a small amount was found in the 0 to 25% fraction.

Aliquots of 10 ml from the 25 to 45% ammonium sulfate fractionations were applied to a fast protein liquid chromatography (Amersham Pharmacia Biotech) MonoQ preparative column and washed with 50 mM Epps buffer as described above. A 0 to 0.5 M KCl gradient (70 ml) was applied, and 2-ml fractions were collected. A 1 M step gradient was used to remove all remaining protein. DAHP synthase activity was pooled from tubes which eluted at about the 0.24 M portion of the KCl gradient. This enzyme fraction was used in most of the assays for characterization.

Antibody preparation.

Regions from the pretransmembrane sequence (periplasmic) and from the posttransmembrane sequence (cytosolic) of X. campestris DAHP synthase were chosen for analysis to determine the best stretch of sequence to use for peptide synthesis and production of antipeptide antibodies. The chosen regions were analyzed by the use of the Network Protein Sequence Analysis at Institut de Biologie et Chimie des Protéines, Lyon, France, to predict antigenic determinants, antigenicity, chain flexibility, and hydrophobic character. After the most favorable sequences for peptide synthesis on each side of the transmembrane region were chosen, they were submitted to Aves Labs, Inc. (Tigard, Oreg.), for further analysis and synthesis. An N-terminal cysteine residue and a spacer amino acid (amino caproic acid) were added to each peptide. These additions allow the use of crude peptide for conjugation purposes and serve to avoid potential problems with steric hindrance.

The synthesized peptides were used to inoculate chickens which lay about one dozen eggs in about 14 days and can produce about 1 g of purified antibody. The highly enriched immunoglobulin fraction from the yolks was then purified (about 500 mg of immunoglobulin Y immune serum antibody). Preimmune serum for use in Western blots and electron microscopy studies was also obtained.

SDS-PAGE and Western blot analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed with a Mini-Protein II cell (Bio-Rad) by using the method of Laemmli (19). Crude extracts were usually diluted 1:1 (vol/vol) with solubilization buffer and heated to 100°C for 10 min. Whole-cell lysis was performed by resuspending the pellet in 2× SDS sample buffer, vortexing vigorously to shear the chromosomal DNA, and heating to 70°C for about 5 min. Samples ranging from 5 to 10 μl were loaded onto SDS–12% polyacrylamide gels. MultiMark (Novex, San Diego, Calif.) multicolored standards were used as protein markers in order to estimate the efficiency of protein transfer and to determine the molecular weight of proteins of interest.

A Trans-Blot SD semidry transfer cell (Bio-Rad) was used to transfer the protein from the mini-gels to nitrocellulose membranes sandwiched between extra-thick blot-absorbent filter paper (Bio-Rad) layers for about 15 to 30 min. Detection of antigens was performed with a chemiluminescence detection system (femtoLUCENT; GENO TECH) according to procedures recommended by the manufacturer.

Cloning of a DNA fragment with DAHP synthase activity.

The strategy followed to clone the gene coding for DAHP synthase from X. campestris was to transform an E. coli strain devoid of DAHP synthase activity with a genomic library from X. campestris, followed by selection for growth in the absence of exogenous l-tryptophan. E. coli strain AB3248 lacks DAHP synthase activity due to mutations that inactivate each of the three genes coding for the differentially regulated paralog isozymes present. This strain is not able to grow in a minimal medium unless all aromatic amino acids and vitamins are added (35).

Chromosomal DNA from X. campestris was partially digested with Sau3A, ligated into the unique BamHI site present downstream of the tac promoter in the expression vector pTacIQ, and transformed into strain XL1 Blue MR. Plasmid DNA purified from this strain was used to transform strain AB3248, which was then plated on ARO minimal medium containing ampicillin (the selecting antibiotic) and IPTG (to induce the tac promoter present on plasmid pTacIQ). From about 30,000 CFU, four colonies emerged on the ARO medium plates after 48 h of incubation at 37°C. To test for reversions of any of the DAHP synthase isozyme mutant genes, plasmid DNA was isolated from each of these four transformants and used to retransform strain AB3248. Only one of the plasmids restored growth of strain AB3248 on minimal medium. This plasmid was named pTacIQ2.3. Restriction analysis of the plasmid revealed the presence of a 6.5-kb DNA segment from X. campestris. Subcloning experiments showed that a 2.7-kb partial BamHI-derived fragment present in plasmid pTacIQ2.3B was still capable of restoring the growth of strain AB3248 in ARO medium.

One possible open reading frame was deduced but had two possible translation start sites (Fig. 1). The largest (M1) was 478 amino acid residues and would have a calculated pI of 5.94 and a molecular mass of 53.4 kDa. The other (M2) would have a calculated pI of 6.14 and a molecular mass of 51.9 kDa. Both of these were subcloned into vector pET-24b⁺ (see Materials and Methods) to generate plasmids pET-M1 and pET-M2. These plasmids were introduced into E. coli BL21(DE3) to obtain clones M1 and M2.

DAHP synthase in crude extracts.

The specific activity of DAHP synthase (AroA_II) in unfractionated extracts of wild-type X. campestris was 5 to 8 nmol/min/mg in 50 mM potassium phosphate buffer at pH 7.0 (36). Distinctly improved specific activities (25.6 ± 1.9 nmol/min/mg) were obtained in 50 mM EPPS buffer at pH 8.3. The pH optimum was broadly similar between pHs of 8.2 and 8.5, dropping sharply under pH 8.2. Significant derepression of AroA_II was not observed when a culture in the mid-exponential stage of growth in rich medium was centrifuged and resuspended in minimal medium for 2 h prior to extract preparation. As another approach, a culture in minimal medium was exposed to 10 mM glyphosate to cause starvation for aromatic amino acids (9). Although growth inhibition was observed, derepression of AroA_II did not occur. Hence, the enzyme may not be subject to repression control by amino acid end products.

Crude extracts prepared from the E. coli BL21(DE3) clones M1 and M2 exhibited elevated specific activities of 796 and 3,109 nmol/min/mg, respectively. Thus, compared to the wild type, the level of AroA_II in the M2 clone was enriched 25-fold or more. A preliminary series of assays comparing the enzymological properties of unfractionated extracts from the wild-type, M1, and M2 strains was undertaken to assess possible differences in catalytic properties, temperature optimum, pH optimum, and divalent cation requirements or in sensitivity to allosteric effectors. The background DAHP synthases that were present in E. coli BL21(DE3) clones were minimized by their repression to low levels in LB and the use of phenylalanine and tyrosine to inhibit the major E. coli paralogs. Characterization of the X. campestris AroA_II protein in the E. coli AB3248 transformant was, of course, uncomplicated by any background of E. coli enzymes. We showed that purified E. coli tryptophan-sensitive DAHP available in our laboratory separated completely from X. campestris AroA_II during the first step of purification. No significant differences between M1 and M2 were found. Multiple alignment results were consistent with the probability that clone M2 possesses the authentic translation start site since the length of clone M2 corresponded best with its nearest relatives, e.g., AroA_II from Helicobacter pylori or from Pseudomonas aeruginosa. The initial conservation observed at the N terminus of clone M2 was 16-WSPQSW-21, which corresponded with 6-WSPTSW-11 of H. pylori and 5-WSPESW-10 of P. aeruginosa. AroA_II from the M2 clone therefore seemed to be the most appropriate protein for use in detailed studies.

Crude extracts from both wild-type X. campestris and from the M1 or M2 clone in E. coli were stable at −70°C for at least 2 months. When aliquots of M2 extract that had been diluted 100-fold in 50 mM EPPS buffer (pH 8.5) were compared with aliquots diluted 100-fold in potassium phosphate buffer at pH 7.5, the latter preparations were significantly more stable in storage at −70°C. In EPPS buffer the half-life was about 4 days compared to about 24 days in phosphate buffer.

Partial purification.

Characterization of AroA_II was carried out in EPPS buffer, the optimum buffer for catalysis. This exacerbated the problem encountered with stability, AroA_II being increasingly unstable with increasing purification or increasing dilution. No stabilizing conditions or agents were found, except for a modest effect of 30% glycerol. However, glycerol was not used because of interference with the chemical assay for DAHP.

In order to deal with the stability problem, a 1-day purification method was worked out for the enzymological characterization of AroA_II from the M2 clone. Unfractionated extract (see Materials and Methods) was subjected to ammonium sulfate fractionation. Ninety-nine percent of the activity salted out in the 25 to 45% (NH₄)₂SO₄ fraction. An assay after resolubilization and dialysis showed an excellent 10-fold purification (266 ± 14 nmol/min/mg). Application to an anion-exchange column yielded fractions which exhibited an additional purification of about ninefold (2.40 ± 0.16 μmol/min/mg). This protocol gave consistent preparations for characterization studies. A single major band was visualized after SDS-PAGE, but multiple minor bands were seen when the gel was overloaded. Western blots confirmed the presence of a single major band of the expected molecular weight. Further attempts to purify the enzyme from gel filtration columns were unsuccessful. Exposure of enzyme preparations, either crude X. campestris extracts or purified protein, to antibody resulted in greater than 99% inhibition of DAHP synthase activity.

Temperature optimum.

Preliminary experiments indicated that AroA_II exhibited a temperature optimum in the range of 50 to 56°C. Detailed assays were carried out in this range to locate the highest temperature at which proportionality was maintained throughout the duration of the 4-min assays used. The temperature optimum was 52°C.

The energy of activation determined according to the Arrhenius equation from a plot of log V (velocity) versus the reciprocal of the temperature (Kelvin scale) was calculated from the negative slope −E/R (E is activation energy and R is the gas constant) and had a value of 11.9 ± 0.9 kcal mol⁻¹.

Effect of DTT.

Since AroA_II from higher plants is hysteretically activated by DTT (7, 11), the possible activation of X. campestris AroA_II by DTT was examined in all strains studied and at different stages of purification. No effect of DTT was found with active preparations. Preparations that had lost some or all of the original activity were also examined, but reactivation by DTT was not found.

Effect of divalent metals.

Neither unfractionated nor partially purified extracts were affected by the presence of divalent metals, as previously noted (36). However, EDTA was a potent inhibitor and was able to completely inactivate the M2 enzyme. Inhibition by EDTA was essentially instantaneous. EDTA was not able to strip essential divalent metal moieties from X. campestris AroA_II, since dialysis of a completely inhibited preparation restored all (107%) of the original activity.

A variety of divalent metals were able to reactivate X. campestris AroA_II (Table 2). Similar results were obtained regardless of whether the enzyme was exposed to EDTA or the divalent metal first or whether EDTA and the test metal were added simultaneously. Co²⁺, Zn²⁺, and Ni²⁺ were the most effective metal additives, at 0.5 mM.

Kinetic properties of X. campestris AroA_II.

A series of substrate saturation curves were generated at different fixed concentrations of each substrate, using elapsed reaction times of 4 min at 37°C. Conditions of proportionality were maintained with respect to reaction time and protein concentration. Replots of data obtained from double-reciprocal plots resulted in apparent K_m values of 0.23 and 0.13 mM for E4P and PEP, respectively.

Analysis of inhibition data confirmed the results of Whitaker et al. (36) that chorismate competitively inhibited each substrate, whereas l-tryptophan was a noncompetitive inhibitor of each substrate. Calculated K_i values for chorismate were 0.31 mM (for PEP) and 0.09 mM (for E4P). Calculated K_i values for l-tryptophan were 0.35 mM (for PEP) and 0.61 mM (for E4P).

Transmembrane region.

When the X. campestris AroA_II sequence was subjected to PSORT analysis (http://psort.nibb.ac.jp), an affirmative result was obtained for a single transmembrane region that would span the inner membrane (Fig. 1). Figure 2 illustrates the 18-amino-acid region predicted to span the inner membrane by the dense alignment surface method. This would place the N-terminal 111 amino acid residues in the periplasm and the C-terminal 349 residues in the cytoplasm. Similar results were obtained for most of the sequences belonging to the upper group shown in Fig. 3. In contrast, members of the lower group were uniformly predicted to be cytoplasmic proteins by application of the PSORT program. The two P. aeruginosa paralogs illustrate these AroA_II differences in a common organism. The biosynthetic pathway protein (upper sequence in Fig. 3) has 448 residues, of which residues 87 to 103 are predicted (certainty of 0.223) to span the inner membrane. The phenazine pathway paralog (lower sequence in Fig. 3) has 405 residues and is predicted to be cytoplasmic, with a certainty of 0.480.

KPRS motif of AroA_II.

Elucidation of the crystal structure of the phenylalanine-sensitive paralog of AroA_I (29) reveals that the crucial amino acid residues which coordinate with metal, PEP, and E4P are scattered throughout the primary sequence between C₆₁ and D₃₂₆. The most compact array of crucial E. coli AroA_I residues is the motif 97-KPRT-100. K₉₇ is bridged by water to a divalent metal and also coordinates with the carboxylate moiety of PEP. R₉₉ is inferred to be an E4P-coordinating residue. The AroA_II family enzymes all share a very similar motif (140-KPRS-143 in X. campestris). The nearby E. coli AroA_I residue R₉₂, which coordinates with PEP, may be equivalent to X. campestris R₁₃₃, which is completely conserved in AroA_II proteins.

Another motif of interest is E. coli AroA_I 265-DxxHxN-270, from which H₂₆₈ coordinates with both PEP and divalent metals. The X. campestris AroA_II motif 374-DxxHxN-379 may be a functional equivalent since this motif is conserved throughout the AroA_II family.

Comparison of X. campestris AroA_II with higher-plant-type AroA_II.

The plastid-localized AroA_II in higher plants (denoted DS-Mn) was the first AroA_II species to be extensively characterized. All higher-plant-type AroA_II proteins share a number of distinctive properties (7, 8, 22). Since the microbial and higher-plant-type AroA_II proteins exhibit relatively high amino acid sequence identities, it was of interest to determine what properties of plant AroA_II might also typify X. campestris AroA_II. Both proteins function in primary aromatic amino acid biosynthesis, and it was shown that in each case (8, 36) a single enzyme species is tightly regulated via an allosteric pattern of control called sequential feedback inhibition (16). However, plant AroA_II and X. campestris AroA_II recognize different key allosteric effectors, l-arogenate and chorismate, respectively. This reflects different biochemical pathway steps diverging from prephenate in these organisms (36). The minor sensitivity of X. campestris AroA_II to inhibition by l-tryptophan is paralleled by a similar sensitivity noted for plant AroA_II in at least one case (mung bean) (26). AroA_II from either source is quite labile and exhibits a relatively high temperature optimum. The putative transmembrane region of X. campestris AroA_II aligned with a hydrophobic region in higher plants, which may also be a transmembrane region. In contrast to the aforementioned similarities, a number of striking differences are also apparent.

Higher-plant-type AroA_II exhibits dramatic hysteretic activation by DTT and is thought to be under redox control influenced by endogenous thioredoxin (11). X. campestris AroA_II was not affected by DTT, and inactive enzyme preparations could not be rejuvenated with DTT. In addition, whereas higher-plant-type AroA_II displays sigmoid substrate saturation curves with respect to either E4P or PEP, X. campestris AroA_II produced straightforward first-order kinetics.

Higher-plant-type AroA_II has activity in the absence of added metal and is stimulated about 50% with the addition of 0.5 mM Mn²⁺. The enzyme is readily stripped free of metal by EDTA treatment to give an inactive preparation which can be restored to original activity levels by Ca²⁺. It has been suggested (8) that two metal-binding sites may exist, one binding either Ca²⁺ or Mn²⁺ and the other binding only Mn²⁺. In contrast, although X. campestris AroA_II is also active in the absence of added metals, further metal additions do not stimulate activity. Thus, X. campestris AroA_II seems to be saturated with firmly bound metal. While EDTA is effective as an inhibitor of activity, it is not able to strip the metal from the enzyme. Mn²⁺ was able to displace the EDTA-metal complex from the X. campestris AroA_II, but Ca²⁺ was ineffective and Co²⁺ and Zn²⁺ were superior to Mn²⁺.

The AroA_II protein is probably conserved throughout the Xanthomonas genus since Xanthomonas hyacinthi (strain ICPB-XH110) was previously reported (13) to possess a DAHP synthase that was 88% inhibited by 0.07 mM chorismate and 33% inhibited by 0.07 mM l-tryptophan.

Independent evolution of tryptophan-inhibited AroA enzymes.

With respect to microbial AroA enzymes that are sensitive to feedback inhibition by l-tryptophan, three classes have been recognized by this criterion (1). First, DS-TRP (exemplified by E. coli) is exclusively subject to inhibition by l-tryptophan. Second, DS-TRP(cha) recognizes l-tryptophan as a major inhibitor and chorismate as a minor inhibitor (exemplified by P. aeruginosa). Third, DS-CHA(trp) recognizes chorismate as a major inhibitor and l-tryptophan as a minor inhibitor. It was proposed (1) that the DS-TRP(cha) gene might have arisen from a fusion between genes encoding an ancestral feedback-insensitive AroA and the large subunit of anthranilate synthase, the latter providing a single source of binding sites for both chorismate and l-tryptophan. Since anthranilate synthase is an aminase, this idea is even more intriguing for the amino-DAHP synthase class of AroA_II (Fig. 3), which introduces an amino group into the reaction (17). However, entry of various AroA_II sequences as queries in PSI-BLAST (National Center for Biotechnology Information) did not return hits for anthranilate synthase.

Since the tryptophan-sensitive E. coli AroA_I protein belongs to a homolog class different from that of tryptophan-sensitive members of AroA_II, it had an independent origin as a paralog member of the AroA_I family. It is interesting that both E. coli and N. crassa possess three differentially controlled isoenzymes of DAHP synthase (14), but whereas E. coli expresses three paralogs of AroA_I, N. crassa expresses two paralogs of AroA_I (phenylalanine and tyrosine regulated) and one of AroA_II (tryptophan regulated). Thus, these classic three-isoenzyme systems, which exhibit a similar functional pattern, differ in that the tryptophan-regulated enzymes had independent evolutionary origins.

AroA_II proteins have a microbial origin.

Figure 3 shows that AroA_II proteins from higher plants are highly conserved and are nested inside a larger and less conserved (more ancient) microbial grouping. The most closely related sequences of the Bacteria include species of Pseudomonas, Campylobacter, Rhodobacter, and Xanthomonas. Although it has been suggested that members of the Bacteria domain might have acquired AroA_II from higher plants (21), the tree shown in Fig. 3 clearly reveals an ancient origin of this AroA family within the Bacteria. Thus, acquisition of AroA_II by higher plants probably occurred relatively recently via endosymbiosis from a donor in the gram-negative lineage of the Bacteria.

Figure 3 shows a clear dichotomy of AroA_II sequences, although this is not necessarily a functional dichotomy. The lower cluster of AroA_II proteins is engaged in phenazine pigment production (8, 23) or antibiotic production. In the latter case, Streptomyces hygroscopicus uses a derivative of DAHP for synthesis of a polyketide starter for rapamycin biosynthesis (25) and Streptomyces collinus utilizes an AroA_II species which produces amino-DAHP as a precursor of 3-amino, 5-hydroxybenzoate, a starter unit for napthomycin (5). These proteins lack the putative transmembrane region, are probably not sensitive to inhibition by chorismate or l-tryptophan, and are encoded by genes belonging to operons whose expression is induced during stationary-phase metabolism. These tend to be the smallest AroA_II proteins, probably because of the deletion of the majority of residues within the region corresponding to X. campestris residues 188 to 243 (Fig. 2). Given the deletion of this region in proteins presumed to lack allosteric control and the variability of this region in organisms recognizing different allosteric effectors, the region corresponding to X. campestris residues 188 to 243 may very well dictate allosteric specificity. (Although the S. collinus sequence is about 100 amino acids longer than other members of the lower cluster, these amino acids are accounted for as N- and C-terminal extensions. It may very well be that the true translational start is a GTG codon that corresponds to residue 68 of the current database entry.)

The upper cluster possesses a number of AroA_II proteins which presumably catalyze the formation of amino-DAHP as a precursor of ansamycin antibiotics (3, 5). Since the AroA_II protein designated Amycolatopsis mediterranei_1 functions as an amino-DAHP synthase, the other, designated A. mediterranei_2, could serve in primary biosynthesis. However, this is not a complete genome, and we note that the AroA_I class of DAHP synthases has been reported in other species of Amycolatopsis (2, 33).

A number of AroA_II proteins in Fig. 3 are from organisms whose genomes have been sequenced. Mycobacterium tuberculosis, Campylobacter jejuni, and H. pylori all possess a single DAHP synthase of the AroA_II type (Fig. 4). It is likely that Corynebacterium diphtheriae (and X. campestris) also possesses a single species of DAHP synthase (i.e., AroA_II). Hence, these AroA_II proteins must function in primary biosynthesis. Particular enzyme reactions can be performed by enzymes which evolved independently and have been referred to as analogous enzymes (10). The frequency of analog proteins which correspond to a given function is surprisingly high (10). It seems quite possible that AroA_II of X. campestris may have originally functioned as an enzyme of secondary metabolism, reminiscent of many contemporary AroA_II enzymes. Chorismate-binding ability might have been a fortuitous property of the enzyme which was subsequently exploited to provide a mechanism of sequential feedback inhibition. Either this facilitated the loss of a less efficient coexisting AroA_I analog, or prior loss of the latter created selective pressure for recruitment of the AroA_II analog to primary biosynthetic function.

Phylogenetic distribution of AroA_II proteins.

If the 16S rRNA tree is accepted as a gauge of phylogenetic breadth, AroA_II clearly exhibits the most narrow phylogenetic distribution of the three groups of AroA proteins (Fig. 4). Thus, it must have the most recent origin. It may have arisen in association with a specialized function for secondary metabolism. Within Mycobacterium, Mycobacterium avium possesses both AroA_Iα and AroA_II, whereas both Mycobacterium tuberculosis and Mycobacterium bovis possess only AroA_II. We suggest that AroA_Iα was the original enzyme serving aromatic biosynthesis in Mycobacterium and has since been replaced by AroA_II. Accordingly, AroA_Iα has been lost in M. tuberculosis and in M. bovis, and it may essentially be an evolutionary remnant in M. avium.

Figure 4 shows that a number of organisms other than X. campestris rely on AroA_II as the sole enzyme available for aromatic amino acid biosynthesis. Although some of these organisms are nutritionally fastidious, the presence of genes specifying complete pathways to aromatic amino acids indicates the presence of a functional pathway. In addition to M. tuberculosis (and M. bovis), these include C. jejuni, H. pylori, Sinorhizobium meliloti, and Legionella pneumophila. Since DAHP synthase is a crucial focal point of allosteric regulation (albeit with remarkable variation of control patterns), one would assume that AroA_II in these organisms is subject to sequential feedback inhibition as in Xanthomonas (via chorismate) or higher plants (via l-arogenate).