Metal-Catalyzed Oxidation of Phenylalanine-Sensitive 3-Deoxy-d-arabino-Heptulosonate-7-Phosphate Synthase from Escherichia coli: Inactivation and Destabilization by Oxidation of Active-Site Cysteines

Abstract

The in vitro instability of the phenylalanine-sensitive 3-deoxy-d-arabino-heptulosonate-7-phosphate synthase [DAHPS(Phe)] from Escherichia coli has been found to be due to a metal-catalyzed oxidation mechanism. DAHPS(Phe) is one of three differentially feedback-regulated isoforms of the enzyme which catalyzes the first step of aromatic biosynthesis, the formation of DAHP from phosphoenolpyruvate and d-erythrose-4-phosphate. The activity of the apoenzyme decayed exponentially, with a half-life of about 1 day at room temperature, and the heterotetramer slowly dissociated to the monomeric state. The enzyme was stabilized by the presence of phosphoenolpyruvate or EDTA, indicating that in the absence of substrate, a trace metal(s) was the inactivating agent. Cu²⁺ and Fe²⁺, but none of the other divalent metals that activate the enzyme, greatly accelerated the rate of inactivation and subunit dissociation. Both anaerobiosis and the addition of catalase significantly reduced Cu²⁺-catalyzed inactivation. In the spontaneously inactivated enzyme, there was a net loss of two of the seven thiols per subunit; this value increased with increasing concentrations of added Cu²⁺. Dithiothreitol completely restored the enzymatic activity and the two lost thiols in the spontaneously inactivated enzyme but was only partially effective in reactivation of the Cu²⁺-inactivated enzyme. Mutant enzymes with conservative replacements at either of the two active-site cysteines, Cys⁶¹ or Cys³²⁸, were insensitive to the metal attack. Peptide mapping of the Cu²⁺-inactivated enzyme revealed a disulfide linkage between these two cysteine residues. All results indicate that DAHPS(Phe) is a metal-catalyzed oxidation system wherein bound substrate protects active-site residues from oxidative attack catalyzed by bound redox metal cofactor. A mechanism of inactivation of DAHPS is proposed that features a metal redox cycle that requires the sequential oxidation of its two active-site cysteines.

The enzyme 3-deoxy-d-arabino-heptulosonate-7-phosphate synthase (DAHPS) (EC 4.2.1.15) catalyzes the condensation of phosphoenolpyruvate (PEP) and d-erythrose-4-phosphate (E4P) to form DAHP and P_i. This reaction is the first committed step in the aromatic biosynthetic pathway in microorganisms and plants, from which tryptophan, tyrosine, phenylalanine, and various aromatic cofactors and secondary metabolites are derived (3, 6). DAHPS is an important control point for metabolic regulation, as it is the earliest pathway target for negative feedback control. In most microorganisms, there are multiple isoforms of DAHPS, which are distinguished by differences in the identities of their specific feedback effectors. In Escherichia coli, there are three DAHPS isozymes, namely, DAHPS(Phe), DAHPS(Trp), and DAHPS(Tyr), each sensitive to feedback inhibition by the respective aromatic amino acid. The three isozymes have a high degree of sequence similarity and undoubtedly have arisen by gene duplication and divergence (20). DAHPS(Tyr) and DAHPS(Trp) are homodimeric enzymes, whereas DAHPS(Phe) is a homotetramer.

The E. coli DAHPS isozymes have in common a requirement for a metal activator that can be satisfied in vitro by a range of divalent metal ions, including Mn²⁺, Fe²⁺, Cd²⁺, Co²⁺, Ni²⁺, Cu²⁺, and Zn²⁺ (27). The identity of the activating metal(s) in vivo has not been established unequivocally, but there is evidence for both Fe²⁺ (15, 19, 27) and Cu²⁺ (2). In vitro activation of the enzyme with Cu²⁺ or Fe²⁺ is accompanied by the appearance of a chromophore with peak A₃₅₀ (19, 27), suggesting a ligand-to-metal charge transfer between bound metal and an enzyme cysteine. There are two invariant cysteines in microbial DAHPS enzymes, namely, Cys⁶¹ and Cys³²⁸ [E. coli DAHPS(Phe) numbering]. Mutational analysis of these two residues in E. coli DAHPS(Phe) (28) has shown that Cys⁶¹ is essential for metal binding and catalytic activity. In contrast, Cys³²⁸ is nonessential, although conservative replacements at this position did have significant negative effects on the kinetic properties of the enzyme, suggesting that it lies near the active site.

It has long been known that the substrate PEP stabilizes DAHPS during purification and storage and protects the enzyme against heat inactivation (14, 17, 22). The second substrate, E4P, has the opposite effect on the stability of DAHPS, increasing the rate of spontaneous inactivation in vitro (15, 17). It has been assumed that this effect is indirect, resulting from the depletion of residual PEP by its enzymatic conversion to DAHP.

Here we show that the in vitro instability of DAHPS(Phe) results from the oxidation of its two invariant cysteines, catalyzed by metal cofactor bound at the metal site, and that PEP, bound at the active site, protects the residues against this attack. All evidence indicates that DAHPS(Phe) is a typical metal-catalyzed oxidation (MCO) enzyme system (24–26).

MATERIALS AND METHODS

Chemicals and enzymes.

1,3-Bis[tris(hydroxymethyl)-methylamino]propane (BTP), PEP (cyclohexylammonium salt), E4P (sodium salt), dithiothreitol (DTT), dimethyl suberimidate, 5,5′-dithiobis-(2-nitrobenzoate) (DTNB), bovine liver catalase, tosylamide phenylethyl chloromethyl ketone (TPCK)-treated trypsin, amino acids, vitamins, and antibiotics were from Sigma Chemical Co. (St. Louis, Mo.); high-purity CaCl₂, CoCl₂, CuSO₄, FeCl₃, FeSO₄, MgCl₂, MnCl₂, and ZnSO₄ were from Aldrich (Milwaukee, Wis.) or Mallinckrodt (Paris, Ky.).

Bacterial plasmids and strains.

The host strain, E. coli CB735 [C600 Δ(gal-aroG-nadA) aroF::Cat^r ΔaroH::Neo^r recA1/F′ lacI^qZΔM15 proA⁺ B⁺ Tn10(Tet^r)], carries a deletion and/or interruption of each of the three chromosomal genes encoding the three DAHPS isoforms (1). Strain CB717 is the recA⁺ version of CB735. Plasmid pTAG1 is a derivative of plasmid pTTG1 (28) in which the transcription of wild-type aroG is driven by one tac and two aroG promoters arranged in tandem. Mutant aroG plasmids pTAG12 (Cys⁶¹→Ser) and pTAG6 (His⁶⁴→Leu) are derivatives of pTAG1, and mutant plasmid pTTG16 (Cys³²⁸→Val) is a derivative of pTTG1; all were constructed by oligonucleotide mutagenesis.

Purification of DAHPS(Phe).

Wild-type DAHPS(Phe) was isolated from overproducing strain CB735/pTAG1. Mutant enzymes were isolated from strains CB735/pTAG6, CB735/pTAG12, and CB717/pTTG16. Growth of cultures, preparation of crude extracts, and purification of DAHPS(Phe) were done as previously described (27). Minor impurities in the final MonoQ preparation were then removed by using hydroxylapatite chromatography as follows. The peak fractions from the MonoQ chromatography were concentrated by ultrafiltration using a Centricell unit (Polysciences Inc., Warrington, Pa.) and loaded on a Bio-Gel HPHT hydroxylapatite column (7.8 mm by 10 cm; Bio-Rad, Rockville Centre, N.Y.) equilibrated with 100 mM sodium phosphate–0.01 mM CaCl₂ (pH 6.8). The column was developed at a flow rate of 0.5 ml/min with a 20-ml sodium phosphate gradient (0 to 1.0 M, pH 6.8). The DAHPS(Phe) peak emerged at approximately 400 mM sodium phosphate. PEP (200 μM) and EDTA (2 mM) were added immediately to the fractions containing DAHPS activity. The combined fractions were concentrated by ultrafiltration using a Centricon 10 unit (Amicon, Beverly, Mass.) and then dialyzed on ice against 20 to 50 volumes of metal-free BPP (20 mM BTP, 200 μM PEP, pH 6.8) with two or three changes. Metal-free BPP buffer was prepared by using Chelex 100 chelating resin (Bio-Rad). Before use, dialysis tubing was boiled in 2% sodium bicarbonate–1 mM EDTA–1 mM EGTA and then exhaustively washed with Chelex-treated water.

The purified enzymes were homogeneous as determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and Coomassie blue staining. The wild-type enzyme had no more than 1 to 2% activity when assayed without metal and was stable for months when stored in small aliquots at −70°C. ApoDAHPS(Phe) (i.e., the PEP- and metal-free enzyme) was prepared immediately before use by gel filtration using a G25 Sephadex PD10 column equilibrated with metal-free BTP buffer (20 mM BTP adjusted to pH 6.8 with HCl).

Determination of protein concentration.

The protein concentration of crude and partially purified preparations of DAHPS(Phe) was determined colorimetrically by the method of Bradford (4) using the Bio-Rad protein assay reagent with bovine serum albumin as the standard. The concentration of purified preparations was determined spectrophotometrically at 280 nm by using a molar extinction coefficient (ɛ^M₂₈₀) of 40,500 for the DAHPS(Phe) monomer (27). The DAHPS(Phe) concentrations given are molar concentrations of the enzyme monomer.

Enzymatic assay.

DAHPS activity was assayed spectrophotometrically (27) by measuring the rate of disappearance of PEP at 25°C. Reaction mixtures contained 20 mM BTP (pH 6.8), 150 μM PEP, 450 μM E4P, and, unless indicated otherwise, 50 μM Mn²⁺. Reactions were generally initiated by addition of 10 to 70 nM enzyme to the reaction mixture.

Determination of histidine and cysteine residues.

Histidine residues of native and inactivated DAHPS(Phe) were titrated by treating the enzyme (3 to 5 μM) in 20 mM phosphate buffer, pH 7.0, with 0.2 mM diethyl pyrocarbonate (Aldrich) at 25°C (16). The increase in A₂₃₀ was monitored until a maximum was attained. The amount of N-carbethoxyhistidyl adduct formed was calculated by using an ɛ^M₂₃₀ of 3,000.

Cysteine residues were titrated with DTNB at 25°C in 20 mM BTP (pH 7.0)–2% SDS as described previously (28). Immediately before analysis, enzyme samples (2 to 4 μM) were applied to a Sephadex G25 PD-10 column (Pharmacia LKB) equilibrated with 20 mM BTP (pH 7.0) and then brought to 2% SDS. After 5 min, DTNB (20 mM in dimethylformamide) was added to 100 μM and the modification reaction was monitored by measuring the increase in A₄₁₂ in a Hewlett-Packard diode array spectrophotometer. The number of modified Cys residues was calculated from the limit absorbance using an ɛ^M₄₁₂ of 14,500 for the released nitromercaptobenzoate anion.

Peptide mapping.

DAHPS(Phe) samples were applied to a Sephadex G25 PD-10 column equilibrated and developed with 100 mM tris(hydroxymethyl)aminomethane-HCl (pH 8.2). The proteins (400 to 800 μg) were digested with TPCK-treated trypsin (20:1, wt/wt) for 24 h at 37°C. The trypsin was added to the reaction mixture in three equal portions at 8-h intervals. The digestion was terminated by addition of 1/50 volume of glacial acetic acid. The tryptic peptides (80 to 100 μg) were fractionated by reverse-phase high-performance liquid chromatography using a Vydac C₁₈ column (0.46 by 25 cm). The column was equilibrated with 5% acetonitrile–0.1% trifluoroacetic acid and developed with a linear 5 to 95% acetonitrile gradient containing 0.1% trifluoroacetic acid at a flow rate of 1%/min. The elution of peptides was monitored by measuring the A₂₁₄. Fractions were dried in vacuo at room temperature in a Savant Speedvac. N-terminal sequencing and mass spectrometry of peptides were performed at the University of Virginia Biomolecular Research Facility.

RESULTS

In vitro stability of DAHPS(Phe).

ApoDAHPS(Phe) displayed significant instability during storage in buffer free of ligands and additives, losing activity steadily over time, with a half-life of about 1.2 days at 22°C (Fig. 1). However, in the presence of 1 mM PEP, the enzyme was very stable, losing little activity after 4 days under the same conditions. It was fortuitously discovered in these experiments that 1 mM EDTA was as effective as PEP in stabilizing the enzyme. Neither PEP nor EDTA, nor a combination of the two, was able to reactivate the partially decayed apoenzyme, but each did prevent further inactivation (data not shown).

FIG. 1.

In vitro stability of DAHPS(Phe). ApoDAHPS(Phe) (∼20 μM) was treated at 22°C with PEP (1 mM) and/or EDTA (1 mM) as indicated. Symbols: ○, no addition; ●, PEP; ▿, EDTA; ▾, PEP plus EDTA.

The structural integrity of the apoenzyme during spontaneous decay was assessed by monitoring its oligomeric state by size exclusion fast protein liquid chromatography (FPLC) (Fig. 2). It was found that the loss of enzymatic activity in the absence of PEP or EDTA was correlated with the slow dissociation of the enzyme (Fig. 2B), which reached completion after 4 days. In contrast, in the presence of PEP and/or EDTA, the enzyme retained its native tetrameric form (Fig. 2A). It was not possible to discern the mode of dissociation; however, since it is known from the crystal stucture that the enzyme tetramer is made up of a loosely associated pair of dimers with tightly associated monomeric subunits (23), it is likely that the dissociation was from tetramer to dimer to monomer. Analysis of the column fractions for DAHPS activity revealed that the tetrameric form of the enzyme was the only active species (data not shown). The dissociation was corroborated by chemical cross-linking of the enzyme preparations with dimethyl suberimidate. Fully decayed preparations showed essentially no cross-linked species on SDS-polyacrylamide gel electrophoresis, whereas fully stabilized and partially inactivated preparations showed cross-linked dimers, trimers, and tetramers, as expected (data not shown).

FIG. 2.

Analysis of subunit dissociation of DAHPS(Phe) by size exclusion FPLC. Enzyme preparations were incubated at 22°C for the indicated times and then fractionated on a Superose 12 column using 20 mM BTP–100 mM KCl (pH 6.8) at a flow rate of 0.5 ml/min. Protein was detected by measuring the A₂₈₀. PEP and EDTA were at 1 mM, as indicated. (A) Apoenzyme incubated with either PEP, EDTA, or PEP plus EDTA. (B) Apoenzyme incubated without addition. The expected elution times of the various oligomeric forms of the enzyme, extrapolated from a standard curve prepared with a set of proteins whose molecular weights are known, were as follows: tetramer, 21.1 min; dimer, 23.3 min; monomer, 25.0 min. The dashed vertical line in panel B (2 days) marks a switchover point for the automated integration of the peaks.

These results indicate that trace metal ions present in the preparation were the primary cause of inactivation and subunit dissociation and that the stabilizing effect of PEP and EDTA was to protect the enzyme from this metal effect. The possibility that PEP acts like EDTA as a metal chelator was discounted by the weak affinity of PEP for metal (K_d^Mn = 2.3 mM) under the conditions used in the stability experiments, as determined by electron paramagnetic resonance analysis (18).

Effects of metals on stability.

In order to ascertain whether the spontaneous inactivation of apoDAHPS(Phe) was metal specific, the enzyme was treated for 18 h at 22°C with a variety of metal activators both in the presence and in the absence of PEP. It was found that 20 μM Cu²⁺ caused complete inactivation of the enzyme and destabilization of its quaternary structure. Fe²⁺ was partially effective, leading to 60% inactivation at 20 μM and 90% inactivation at 200 μM. Co²⁺, Mn²⁺, Zn²⁺, and Fe³⁺ had little or no effect, even at 10-fold higher concentrations. PEP (400 μM) afforded full protection from inactivation by 20 μM Cu²⁺ and Fe²⁺ and partial protection at 200 μM Cu²⁺. The lesser effects of Fe²⁺ were apparently due to the oxidation of Fe²⁺ to Fe³⁺ under the aerobic conditions of the experiment, since the addition of freshly prepared Fe²⁺ to the partially inactivated enzyme led to complete inactivation and dissociation. In light of this, Cu²⁺ was chosen for further characterization of the phenomenon of metal inactivation.

The rate of Cu²⁺-dependent inactivation was found to be time and concentration dependent (Fig. 3). The half-life of inactivation at 22°C in the absence of PEP was about 4 h with 20 μM Cu²⁺ (approximately equimolar with the enzyme) and 1 h with 200 μM Cu²⁺ (10-fold molar excess). As before, PEP afforded protection at both concentrations. Size exclusion FPLC fractionation of the Cu²⁺-inactivated apoenzyme having ∼10% residual activity, obtained after 3 h of incubation with 200 μM Cu²⁺, revealed only partial dissociation of the enzyme, with about two-thirds of the molecules remaining in the tetrameric state and the other third distributed between the dimeric and monomeric forms (data not shown). However, upon continued incubation, dissociation of the enzyme eventually reached completion at both concentrations of Cu²⁺. Thus, the loss of enzymatic activity preceded the dissociation of the enzyme to its subunits.

FIG. 3.

Inactivation of DAHPS(Phe) by Cu²⁺. ApoDAHPS(Phe) (∼20 μM) was treated at 22°C with the indicated concentrations of CuSO₄ in the presence or absence of 500 μM PEP. Residual enzymatic activity was measured at the indicated times. Symbols: ○, no Cu²⁺ but no PEP; ●, PEP but no Cu²⁺; ▿, 20 μM Cu²⁺ but no PEP; ▾, 20 μM Cu²⁺ plus PEP; □, 200 μM Cu²⁺ but no PEP; ■, 200 μM Cu²⁺ plus PEP.

The results of these experiments indicate that inactivation and destabilization of DAHPS(Phe) occurred when one of the redox metal cofactors, copper or iron, was bound to the enzyme in the absence of bound PEP. Support for this conclusion was found in the stability properties of mutant forms of the enzyme with impairments in the binding of either metal or PEP. Previous mutational studies of DAHPS(Phe) have established that the highly conserved sequence Gly⁵⁹ProCysSerIleHisAsp⁶⁵ is an essential component of the active site of the enzyme (18). Mutant forms of the enzyme with conservative changes at Cys⁶¹ no longer bind metal (K_d^Mn, >200-fold that of the wild-type enzyme) and have no catalytic activity. In contrast, mutant forms with changes at His⁶⁴ have reduced affinity for PEP (i.e., increased K_d^PEP) but retain metal binding and enzymatic activities. It was found that under conditions in which the wild-type apoenzyme was rapidly and completely dissociated to monomeric subunits (200 μM Cu²⁺, no PEP), the inactive Cys⁶¹→Ser enzyme remained completely tetrameric (data not shown). Conversely, under conditions in which the wild-type enzyme was stable for >6 days (400 μM PEP), the activity of the His⁶⁴→Leu mutant enzyme, whose K_d^PEP is ∼15-fold greater than that of the wild-type enzyme (18), decayed with a half-life of about 24 h.

These stability properties of DAHPS(Phe) are typical of MCO enzyme systems (25, 26). In MCO systems, amino acid residues at the active site become susceptible to oxidative modification catalyzed by bound metal cofactors in the absence of bound substrate. In many MCO enzymes, protection from oxidation can be achieved in the absence of bound substrate by exclusion of molecular oxygen or by elimination of oxidative intermediates, such as H₂O₂ (25). Accordingly, the effect of anaerobiosis and of catalase on the inactivation of apoDAHPS(Phe) by Cu²⁺ was examined. As shown in Fig. 4, both treatments significantly reduced the rate of inactivation.

FIG. 4.

Effect of catalase and anaerobiosis on inactivation of DAHPS(Phe) by Cu²⁺. ApoDAHPS(Phe) (13 μM) was treated with 200 μM CuSO₄. Where indicated, PEP (400 μM) or bovine liver catalase (34,000 U) was added 5 min before addition of CuSO₄. Anaerobiosis was achieved by continuous bubbling of N₂ gas through the enzyme solution in a sealed tube. Bubbling was begun 5 min before the addition of CuSO₄. Symbols: ○, no treatment; ●, PEP; ■, catalase; ▾, anaerobiosis.

Identification of target residues.

A variety of active-site residues have been found to be the target of oxidative attack in different MCO enzyme systems, including arginine, cysteine, histidine, lysine, and proline (25). In preliminary attempts to identify the affected residue(s) in DAHPS(Phe), it was found that the rate of spontaneous inactivation of apoDAHPS(Phe) was markedly influenced by pH, increasing as the pH decreased from 7.0 to 6.0. When first-order rate constants of the loss of enzymatic activity were plotted as a function of pH, the derived curve indicated a critical target residue whose pK was ∼6.3, suggesting that the protonated form of histidine is the target for oxidation. However, when the number of histidine residues in spontaneously inactivated apoDAHPS(Phe) was assessed by diethyl pyrocarbonate modification, no significant reduction from the 11 residues present in the native enzyme was found. The apoenzyme treated with 200 μM Cu²⁺ showed a gradual loss of histidine residues with time, increasing from one residue lost after 5 h, at which time there was >90% inactivation, to three residues lost after 24 h. However, the same results were obtained when the metal treatment was carried out with the wild-type enzyme in the presence of 400 μM PEP, where there was little (<10%) loss of enzymatic activity, as well as with the Cys⁶¹→Ser enzyme, which is defective in metal binding. Thus, the copper-catalyzed attack of histidine residues was nonspecific rather than targeted to a residue(s) at the active site of the enzyme.

On the other hand, when the cysteine content of native and inactivated DAHPS(Phe) was analyzed by DTNB modification, significant reductions were found in the inactivated enzymes (Table 1). The spontaneously inactivated apoenzyme (<5% residual activity) showed a net loss of two of the seven cysteine thiols per subunit. The loss was greater in the Cu²⁺-inactivated preparations, increasing with increasing concentrations of added metal. At 10 μM Cu²⁺ (metal/enzyme monomer molar ratio of 0.25), there was a net loss of four thiols per subunit, whereas at both 40 and 400 μM Cu²⁺, the loss was six thiols. PEP was effective in stabilizing the enzyme in the presence of even the highest concentration of Cu²⁺, where the loss in activity was only 17%, even though there was still a net loss of three thiols. These results suggest that PEP preferentially protects an essential cysteine residue(s) at the active site.

TABLE 1.

Thiol content of Cu²⁺-inactivated and DTT-reactivated DAHPS(Phe)^a

Enzyme treatment			DAHPS activityb	Thiols/subunit
Cu2+ concn (μM)	PEP	DTT
None	−	−	2	5.0 ± 0.3
	−	+	107	7.3 ± 0.3
10	−	−	12	2.9 ± 0.2
	−	+	81	6.4 ± 0.1
40	−	−	12	1.2 ± 0.2
	−	+	60	5.4 ± 0.3
400	−	−	11	1.0 ± 0.2
	−	+	37	4.1 ± 0.3
400	+	−	83	4.2 ± 0.1
	+	+	84	5.4 ± 0.2

^aFreshly prepared apoDAHPS(Phe) (40 μM) was treated with various concentrations of CuSO₄ at 22°C until the activity had decayed to around 10%, at which time EDTA (1 mM) was added to terminate the treatment by chelation of the metal. Where indicated, PEP (1 mM) was added immediately before the metal and the preparation was incubated for the same time as the one without PEP. Where indicated, DTT (1 mM) was added after the metal treatment and the preparation was incubated for an additional 5 h. Thiol content was determined by DTNB titration as described in Materials and Methods. 

^bExpressed relative to the initial activity of the untreated enzyme. All activities were determined by using the standard assay procedure with Mn²⁺ as the metal activator. 

Treatment of the spontaneously inactivated apoenzyme with DTT restored the two lost thiols and completely restored enzymatic activity (Table 1). Reactivation was slow, requiring about 5 h to reach completion. Only partial reactivation and thiol restoration was observed upon DTT treatment of the Cu²⁺-inactivated preparations. The extent of reactivation was inversely related to the concentration of Cu²⁺ used for inactivation, indicating that high concentrations of Cu²⁺ cause additional deleterious modifications that are not reversed by DTT reduction. Size exclusion FPLC showed that in all cases, DTT reactivation was accompanied by reassembly of the dissociated subunits of the inactivated enzyme to the tetrameric state in amounts equivalent to the restoration of activity (data not shown).

Identification of oxidized cysteine residues.

In order to identify the oxidized cysteine residues in the inactivated enzyme, peptide mapping of the native enzyme, the spontaneously inactivated enzyme and the Cu²⁺-inactivated enzyme was carried out. The peptide profiles of the native enzyme (Fig. 5A) and the spontaneously inactivated enzyme (Fig. 5C) were similar, although there were quantitative differences in a number of the peaks. In contrast, in the Cu²⁺-inactivated preparation, a major new peak was observed, eluting late in the fractionation (Fig. 5B). In experiments in which a range of Cu²⁺ concentrations (0.25-, 1.0-, and 10-fold molar ratios) was used for inactivation, the novel peptide increased in amount as the concentration of Cu²⁺ increased. In all cases, DTT reactivation of the Cu²⁺-inactivated enzyme preparations led to the disappearance of the novel peptide (data not shown). It was not possible to correlate the absence of a specific peak in the peptide profile of the native enzyme with the appearance of the novel peak, perhaps because of ɛ^M₂₁₄ differences in the peptides.

FIG. 5.

Tryptic peptide maps of native and inactivated DAHPS(Phe). Peptide mapping was carried out as described in Materials and Methods. Panels: A, native apoDAHPS(Phe) (fully active); B, apoDAHPS(Phe) treated with 400 μM CuSO₄ for 4 h at 22°C (>90% inactivation); C, apoDAHPS(Phe) incubated for 4 days at 22°C (>90% inactivation). The arrow in panel B indicates the novel peptide peak that appears upon Cu²⁺ treatment.

The novel peptide was recovered by fractionation of the Cu²⁺-inactivated enzyme and characterized. Mass spectrometry analysis showed a single peptide species with a mass of 3,605, a value greater than that of the largest tryptic peptide expected from the native enzyme. However, amino-terminal sequencing revealed that there were two equimolar amino termini in the peptide sample, indicating that the novel peak was composed of two peptides that were covalently linked. The results of six cycles of sequencing unequivocally identified the two cross-linked peptides as L⁵⁴LVVIGPC⁶¹SIHDPVAAK⁷⁰ and S³²³ITDAC³²⁸IGWEDTDALLR³³⁹. Significantly, each of the peptides contained one of the enzyme’s two active-site cysteines, Cys⁶¹ and Cys³²⁸. The calculated combined mass of these two peptides was 3,608, nearly identical to that determined for the isolated dipeptide. These results established that the novel peptide in the Cu²⁺-inactivated enzyme was composed of the two peptides oxidatively cross-linked by a disulfide bridge between Cys⁶¹ and Cys³²⁸.

It is noteworthy that the cross-linked dipeptide was not present in the tryptic map of the spontaneously inactivated enzyme (Fig. 5C), indicating that oxidative attack of the two cysteines and inactivation of the enzyme do not necessarily involve formation of the disulfide bridge between the two target cysteines. Nevertheless, it was found that treatment of the spontaneously inactivated enzyme with a 10-fold molar excess of Cu²⁺ before digestion with trypsin led to the appearance of the dipeptide peak in the fractionation (data not shown).

Stability of Cys³²⁸→Val DAHPS(Phe).

As mentioned above, the inactive Cys⁶¹→Ser enzyme, which does not bind metal, was found to be resistant to metal attack, as judged by the lack of subunit dissociation upon Cu²⁺ treatment. This raised the question of whether a mutational change at Cys³²⁸ would also affect metal sensitivity. Accordingly, the stability of the Cys³²⁸→Val enzyme, previously shown to have only slightly impaired catalytic properties, i.e., a 20% reduction in the catalytic constant and two- to threefold increases in K_m^PEP and K_m^E4P (28), was examined. It was found that the Cys³²⁸→Val apoenzyme was completely resistant to both spontaneous and Cu²⁺-catalyzed inactivation (Fig. 6), demonstrating that metal attack of DAHPS(Phe) requires the presence of both the Cys⁶¹ thiol and the Cys³²⁸ thiol.

FIG. 6.

In vitro stability of Cys³²⁸→Val mutant DAHPS(Phe). ApoDAHPS(Phe) (∼20 μM) was treated at 22°C as indicated. Symbols: ○, ●, wild-type enzyme; ▿, ▾, Cys³²⁸→Val enzyme. Open symbols, no added Cu²⁺ (days timescale); filled symbols, 250 μM CuSO₄ (hours timescale).

DISCUSSION

The results reported here establish that the in vitro instability of DAHPS(Phe) in the absence of bound PEP is due to the oxidation of two active-site cysteinyl residues, Cys⁶¹ and Cys³²⁸, catalyzed by redox metal ions that normally activate the enzyme. The oxidation of thiol groups in proteins proceeds by a succession of single-electron transfers, giving rise to a progression of oxidation states (Fig. 7), all of which, except for the sulfonic acid, are reversible by DTT and other reducing agents (10, 13).

FIG. 7.

Succession of single-electron transfers giving rise to a progression of oxidation states of the thiol group.

The first two derivatives of the oxidative pathway, the thiyl radical and the sulfenic acid, are highly reactive, and if two such adducts exist in close enough proximity and are sterically free to interact, they will spontaneously form a disulfide linkage (12, 13). Thus, it is possible that during spontaneous decay of DAHPS(Phe) (i.e., trace amounts of metal ions), the thiyl or sulfenic acid derivatives of Cys⁶¹ and Cys³²⁸ are formed and, perhaps because of steric limitations and the protective environment of the active site, are able to persist as such. This is consistent with the crystal structure of the enzyme, where it has been found that Cys⁶¹ and Cys³²⁸ are in close proximity within the active site (Cα’s within ∼6 Å), but their two thiol groups are not favorably positioned for disulfide bond formation (23). For the thiols to attain optimal orientation, the side chain of Cys⁶¹ must undergo a net rotation of about 180° about its Cα-Cβ bond. Thus, under conditions of high metal concentration, significant conformational changes must accompany the oxidation of the two cysteine thiols, permitting formation of the disulfide bridge.

Besides metal and O₂, MCO systems characteristically require either a nonenzymatic electron donor such as ascorbate, NAD(P)H, or thiol compounds or an enzymatic system for the generation of needed electrons (24, 25). The electron donor plays a dual role in the reduction of Cu²⁺ or Fe³⁺ to Cu¹⁺ or Fe²⁺ and in the production of H₂O₂ by the reduction of O₂. The H₂O₂ then disproportionates via Fenton chemistry to a hydroxyl radical (.OH), the reactive oxygen species ultimately responsible for oxidative damage to proteins, and a hydroxyl ion (OH⁻) with the regeneration of the oxidized metal ion (5, 25). Since the DAHPS(Phe) system does not require an exogenous electron donor, the electrons required for H₂O₂ formation must be derived endogenously. A possibility that is consistent with the data is that the thiol group of Cys³²⁸ serves as the required electron donor.

Based on this assumption, the following mechanistic model for the DAHPS(Phe) MCO system is postulated: (1) Cys³²⁸-SH + Cu²⁺ + O₂ → Cys³²⁸-S. + Cu¹⁺ + H₂O₂(2) Cu¹⁺ + H₂O₂ → Cu²⁺ + .OH + OH⁻(3) Cys⁶¹-SH + .OH → Cys⁶¹-S. + H₂O(4) Cys⁶¹-S. + Cys³²⁸-S. → Cys⁶¹-S-S-Cys³²⁸The sequence is initiated by the reduction of O₂ to H₂O₂ with the concomitant oxidation of the thiol group of Cys³²⁸ to the thiyl radical, catalyzed by redox metal present at the active site (reaction 1). The reduced Cu¹⁺ ion formed in reaction 1 then catalyzes the formation of the hydroxyl radical from H₂O₂ via Fenton chemistry, regenerating Cu²⁺ (reaction 2). The hydroxyl radical, in turn, attacks Cys⁶¹, oxidizing its thiol group to the thiyl radical (reaction 3), thereby eliminating metal binding and inactivating the enzyme. In the spontaneously inactivated enzyme, the two oxidized cysteine thiols remain unbridged. However, in the Cu²⁺-inactivated enzyme, additional oxidative modifications of the enzyme occur, such as the attacks on the nonessential cysteine and histidine residues that were detected (Table 1), that lead to the conformational changes necessary for the repositioning of the Cys⁶¹ side chain and spontaneous formation of the disulfide (reaction 4). It is likely that the same or similar modifications occur in the spontaneously inactivated enzyme posttreated with Cu²⁺, since this treatment triggers disulfide formation in the already inactive enzyme. These nonspecific modifications presumably derive from the ability of the metal in solution to act as a free radical in the generation of activated oxygen derivatives (5) and thus would not be restricted to active-site residues.

The first three steps of the MCO model described above envisions a “caged” reaction at the active site of DAHPS(Phe) that involves the sequential oxidation of Cys³²⁸ and Cys⁶¹ by distinctly different mechanisms. It explains why modification of the nonessential Cys³²⁸ residue is required for inactivation of the enzyme (Fig. 6). It also explains how trace amounts of metal are able to accomplish the slow inactivation of a large excess of enzyme molecules (Fig. 1). Reactions 1 and 2 constitute a redox cycle that regenerates Cu²⁺ which, no longer coordinated by the oxidized thiol of Cys⁶¹, is free to exit the damaged active site and diffuse to another intact active site to repeat the cycle. It is postulated that PEP bound at the active site blocks initiation of the cycle by sequestering the bound metal, thereby protecting both Cys³²⁸ and Cys⁶¹ from oxidative attack. This is consistent with both crystallographic and ligand binding studies of the enzyme. The crystal structure of DAHPS(Phe) has revealed that the metal is coordinated by bound PEP at the active site (23), an interaction that increases the binding constant of both ligands by an order of magnitude (18). On the other hand, no insight can be derived from the crystal structure of the enzyme to explain how oxidation of the two active-site cysteines leads to subsequent destabilization of the quaternary structure of the enzyme.

In view of the findings that DAHPS(Trp) and DAHPS(Tyr) are also markedly unstable in vitro in the absence of PEP (19, 22), that the three isozymes share the same spectrum of metal activators (27), and that Cys⁶¹ and Cys³²⁸ are invariant in the three isozymes (20), it appears likely that the mechanism of metal-catalyzed oxidation found here in DAHPS(Phe) is operable in the other two isozymes as well. On the other hand, it is not clear whether metal-catalyzed oxidation plays a role in the metabolic regulation of the level of DAHP synthase activity in vivo, as has been shown in other MCO enzyme systems (24, 26). Such a mechanism would imply that the intracellular availability of PEP during growth modulates the susceptibility of the enzyme to metal-catalyzed attack and subsequent proteolytic turnover (8, 9, 21). That iron and/or copper are most likely the activating metals in vivo (2, 19, 27) is consistent with this possibility. However, little is known about the intracellular turnover of the DAHP synthase isozymes in E. coli. In one report, it was proposed that specific proteolytic degradation acts to control the level of DAHP synthase activity during growth (7). This was based on the observation that the activities of the coordinately expressed enzymes DAHPS(Tyr) and chorismate mutase became disproportionate when the cells entered the stationary phase of growth, apparently as a result of the inactivation of DAHPS(Tyr). Thus, it is possible that in the late stages of the growth of E. coli, the normally abundant intracellular pool of PEP (11) becomes depleted, leading to the metal-catalyzed oxidation and degradation of one or more of the DAHPS isozymes. A closer examination of individual isozyme turnover rates, PEP pool levels, metabolite compartmentalization, and metal utilization throughout the growth cycle is needed before the efficacy of this type of metabolic regulation can be established.