Biochemical characterization of prephenate dehydrogenase from the hyperthermophilic bacterium Aquifex aeolicus

Abstract

A monofunctional prephenate dehydrogenase (PD) from Aquifex aeolicus was expressed as a His-tagged protein in Escherichia coli and was purified by nickel affinity chromatography allowing the first biochemical and biophysical characterization of a thermostable PD. A. aeolicus PD is susceptible to proteolysis. In this report, the properties of the full-length PD are compared with one of these products, an N-terminally truncated protein variant (Δ19PD) also expressed recombinantly in E. coli. Both forms are dimeric and show maximum activity at 95°C or higher. Δ19PD is more sensitive to temperature effects yielding a half-life of 55 min at 95°C versus 2 h for PD, and values of k_cat and K_m for prephenate, which are twice those determined for PD at 80°C. Low concentrations of guanidine-HCl activate enzyme activity, but at higher concentrations activity is lost concomitant with a multi-state pathway of denaturation that proceeds through unfolding of the dimer, oligomerization, then unfolding of monomers. Measurements of steady-state fluorescence intensity and its quenching by acrylamide in the presence of Gdn-HCl suggest that, of the two tryptophan residues per monomer, one is buried in a hydrophobic pocket and does not become solvent exposed until the protein unfolds, while the less buried tryptophan is at the active site. Tyrosine is a feedback inhibitor of PD activity over a wide temperature range and enhances the cooperativity between subunits in the binding of prephenate. Properties of this thermostable PD are compared and contrasted with those of E. coli chorismate mutase-prephenate dehydrogenase and other mesophilic homologs.

Prephenate dehydrogenase (PD) is a member of the TyrA protein family involved in the biosynthesis of L-tyrosine. This enzyme catalyzes the oxidative decarboxylation of prephenate to 4-hydroxyphenylpyruvate (HPP) in the presence of NAD⁺ (Fig. 1). This conversion, along with the rearrangement of chorismate to prephenate catalyzed by chorismate mutase (CM), constitutes two consecutive reactions that are essential for tyrosine biosynthesis in many bacteria and other microorganisms, yeast, and fungi (Dayan and Sprinson 1971; Ahmad and Jensen 1988; Ahmad et al. 1990; Schnappauf et al. 1998). In the enteric bacterium Escherichia coli, the two activities are associated with a bifunctional enzyme CM-PD, also known as the T-protein, while in many other organisms the reactions are catalyzed on two separate polypeptides (Koch et al. 1971; Schnappauf et al. 1998). Only a few monofunctional TyrA proteins have been studied in purified form. These include dehydrogenases from Zymomonas mobilis (Zhao et al. 1993), Pseudomonas (Xia and Jensen 1990; Xie et al. 2000), Synechocystis sp. PCC 6803 (Bonner et al. 2004), Arabidopsis thaliana (Rippert and Matringe 2002a), and Neisseria gonorrhoeae (Subramaniam et al. 1994). These studies have centered mainly on their substrate specificity, as some utilize both L-arogenate and prephenate as substrate and others use only L-arogenate. In this alternative pathway, prephenate's side-chain keto group is transaminated to yield L-arogenate, which then undergoes oxidative decarboxylation to L-tyrosine. Most detailed mechanistic studies on prephenate dehydrogenase activity, however, have used the bifunctional E. coli CM-PD. This enzyme is a homodimer of 373 amino acids per monomer and houses the dehydrogenase domain in the latter two-thirds of the polypeptide chain (Hudson and Davidson 1984). Initial velocity, product, and dead-end inhibition studies show that the dehydrogenase conforms to a rapid equilibrium, random kinetic mechanism with catalysis as the rate-limiting step (Sampathkumar and Morrison 1982). A model for the reaction catalyzed by the prephenate dehydrogenase activity of E. coli CM-PD has been developed based on results from the analysis of chemical modification studies (Christendat and Turnbull 1996), pH rate profiles (Turnbull et al. 1991a; Christendat et al. 1998), isotope effects (Hermes et al. 1984), and patterns of inhibition by substrate analogs for wild-type and variant forms of the bifunctional enzyme (Christendat and Turnbull 1999). In this model, Arg294 likely interacts with the ring carboxylate of prephenate, while His197 is believed to polarize the 4-hydroxyl group of prephenate, facilitating hydride transfer to NAD⁺ and concomitant decarboxylation. Kinetic analysis suggests that E. coli CM-PD is allosterically inhibited by tyrosine, although the exact mechanism of inhibition remains unclear. The results from some biophysical studies have led to the interpretation that the enzyme interconverts from an active dimer to an inactive tetramer upon the binding of tyrosine plus NAD⁺ (Hudson et al. 1983), while modeling of kinetic data has allowed for only tertiary structural changes within dimeric PD to elicit tyrosine's effects (Christopherson and Morrison 1985). Some proteins within the TyrA family have been reported to be insensitive to tyrosine inhibition (Zhao et al. 1993).

Figure 1.

Reaction catalyzed by chorismate mutase and prephenate dehydrogenase. Chorismate mutase catalyzes the rearrangement of chorismate (1) to prephenate (2), while prephenate dehydrogenase catalyzes the oxidative decarboxylation of prephenate to (4-hydroxyphenyl)pyruvate (3) in the presence of NAD⁺.

There have been efforts to separate the activities of the T-protein into discrete monofunctional domains. Jensen and coworkers (Xia et al. 1992b) initially reported the successful expression of a PD derived from Erwinia herbicola CM-PD but only when a large portion of the mutase domain was not removed. More recent work by Ganem (Chen et al. 2003) and by our lab (J. Bonvin, unpubl.) showed that independently expressed CM and PD domains of the E. coli enzyme have reduced activity and are highly unstable or insoluble, highlighting the structural interrelationship of the different regions of the polypeptide chain. Only limited properties were reported for the monofunctional PD. In contrast, the related bifunctional enzyme chorismate mutase-prephenate dehydratase (CM-PDT) involved in phenylalanine biosynthesis has been shown to possess two distinct noninteracting catalytic sites (Duggleby et al. 1978; Stewart et al. 1990; Lee et al. 1995; Zhang et al. 1998) and a phenylalanine-binding domain at the C-terminal region of the protein (Zhang et al. 1998; Pohnert et al. 1999), all of which can be separately expressed and are fully functional.

There has been no reported crystal structure of prephenate dehydrogenase or TyrA protein from any organism. As steps toward obtaining structural information on PDs and to offer more insight into its biochemical properties, we initiated studies on a monofunctional PD from the hyperthermophilic bacterium Aquifex aeolicus. A. aeolicus is a chemolithoautotroph that thrives at temperatures >85°C in environments containing only inorganic components and utilizes as substrates gaseous hydrogen, carbon dioxide, and oxygen (Kawasumi et al. 1984; Huber et al. 1992). The complete genomic sequence of A. aeolicus was reported in 1998 from samples isolated from hydrothermal vents in Yellowstone National Park (Deckert et al. 1998). A tyrA gene encoding a 311-residue PD (E.C. 1.3.1.12) was based on its 36% nucleotide sequence identity to known proteins denoted as prephenate dehydrogenases in the GenBank nonredundant database (Deckert et al. 1998). In this report we describe the cloning and heterologous expression of this PD and characterization of its biochemical and biophysical properties. The N terminus of the enzyme is susceptible to proteolysis, and hence we also compare its properties with those of an N-terminal deletion construct, which has yielded diffraction quality crystals.

Results

Protein purification and subunit composition

The tyrA gene from A. aeolicus was cloned into pET-15b. Protein derived from this construct contained a 20-residue N-terminal extension that included a hexa-His tag to facilitate purification by Ni-NTA affinity chromatography and a thrombin cleavage site. Results of the purification are summarized in Table 1 and Figure 2 and showed that both heat treatment and affinity chromatography were effective purification steps. However, the purified His-tagged PD gave two distinct bands on SDS-PAGE (Fig. 2, lane 7), representing full-length (37 kDa) and shortened (34 kDa) forms of the protein. The protein's chromatographic behavior indicated that the 34-kDa subunit lacked the His tag and that native PD was dimeric. As expected, when both subunits of the dimer were missing the tag, the enzyme did not bind to the nickel resin (Fig. 2, lane 5); the flowthrough contained 30% of the total activity of the heat-treated cell extract, indicating that this enzyme form was active. Heterodimeric PD (a monomer each of full-length His-tagged and shortened nontagged forms) was eluted by washing the affinity column with 30 mM imidazole-containing buffer (Fig. 2, lane 6), while the 300 mM imidazole eluate was enriched in homodimeric PD (both subunits tagged) (Fig. 2, lane 7). Treatment of the sample from lane 7 with thrombin yielded protein that appeared as a single band (>98% pure) by Coomassie staining (Fig. 2, lane 8). Proteins in the different purification fractions resolved by SDS-PAGE were also stained with a fluorescent NTA probe that interacts specifically with oligo-His motifs. As predicted, the label was traced to protein in fractions containing the His-tagged 37 kDa subunit of the heterodimer (data not shown).

Table 1.

Purification of A. aeolicus PD expressed in E. coli^a

Figure 2.

SDS-PAGE analysis of the purification of A. aeolicus PD. (Lane 1) Molecular weight standards, (lane 2) cell-free extract (65 μg), (lane 3) heat-treated soluble fractions (10 μg), (lane 4) Q-Sepharose fraction (10 μg), (lane 5) Ni-NTA column flowthrough (5 μg), (lane 6) 30 mM imidazole wash (2 μg), (lane 7) 300 mM imidazole wash (8 μg), (lane 8) thrombin-treated PD (6 μg), (lane 9) thrombin-treated Δ19PD (8 μg).

Development of a reliable protocol for ESI-MS analysis of PD confirmed the identity of each protein. In addition, the results supported the idea that the shortened form was produced by cleavage at the N-terminal region of PD. For clarity, amino acids are numbered for PD from 1 to 311, beginning with PD's authentic Met as annotated from the database; the residues of the tag are not numbered. For heterodimeric PD, analysis showed that the His-tagged subunit lacked the first Met at the tag's N terminus (expected [M + H⁺], 36,881.6 Da; observed, 36,880.5 Da), and that the smaller 34-kDa subunit commenced at Ser5 in PD (expected [M + H⁺], 34,419.9 Da; observed, 34,419.5 Da). Results for the purified thrombin-treated PD verified cleavage at the thrombin-recognition site, between Arg and Gly within the N-terminal tag (expected [M + H⁺], 35,130.8 Da; observed, 35,130.6 Da). The sample also contained the 34-kDa product as a minor species (∼15%). The shortened form of the recombinant PD was not produced by translation starting at positions 1, 7, 30, 38, or 41 of PD (the latter two would result in truncation of the NAD⁺-binding domain), or by cleavage at the thrombin-recognition site. Moreover, none of the peaks resolved by ESI-MS matched mass values predicted for cleavage within the protein's C-terminal region.

We found that purified thrombin-treated PD (1–10 mg/mL) remained stable and fully active when kept at −86°C but was susceptible to further N-terminal degradation if stored for extended periods of time at 4°C or if purified in the absence of insufficient amounts of protease inhibitors. Moreover, thrombin-treated PD subjected to crystallization trials during the course of these studies yielded crystals. The observed susceptibility of PD to nonspecific proteolysis implies that residues at the N-terminal region of PD are exposed and/or within regions of secondary structure that adopt a mostly random conformation (Supplemental Fig. 1). Alternatively, as the N terminus of PD expressed in the native organism has not been demonstrated directly, it is possible that the authentic start site that can translate functional protein is actually located further downstream (i.e., Met30) than that annotated in the database. If so, the sensitive proteolytic targets might belong to an unstructured N-terminal extension of PD. As a result of these observations, several protein variants were expressed, and one of them (Δ19PD) has yielded crystals diffracting to high resolution (Sun et al. 2006). The expression and purification of Δ19PD were performed as described for PD with minor modifications (see Materials and Methods). The yield of Δ19PD is routinely ∼14 mg/L of culture, and neither the His-tagged nor thrombin-treated forms showed heterogeneity during purification. The mass of Δ19PD was verified by ESI-MS (expected [M + H⁺] of His-tagged and thrombin-treated forms, 35,078.5 Da and 33,196.5 Da, respectively; observed, 35,077.1 Da and 33,196.0 Da, respectively) and was checked routinely throughout our studies. Some studies presented in this report have been performed only on PD, while others, on both PD and Δ19PD. Detailed spectroscopic studies were pursued on the homogeneous, crystallizable Δ19PD form.

Native molecular weight

Analysis of PD and Δ19PD by size exclusion FPLC from 0.1 to 2 mg/mL yielded values of 54 kDa and 52 kDa, respectively. These values were less than those predicted for dimeric forms, ∼70 kDa and ∼66 kDa. However, masses calculated from the primary sequence more closely matched those obtained using sedimentation velocity analytical ultracentrifugation. Average molecular masses of 67.9 kDa and 62.5 kDa and sedimentation coefficients S_20,w of 4.72 and 4.90 were calculated for PD and Δ19PD, respectively. Our results were consistent with the thermophilic PDs being dimeric in solution but more compact than the commercially available globular proteins used to calibrate size exclusion columns (Jaenicke 2000). Occasionally our PD preparations contained a catalytically active oligomer of ∼160 kDa (10% of the sample peak on FPLC), that we attributed, in part, to disulfide linked subunits (there is one cysteine/monomer). Heterogeneity was removed by size exclusion chromatography prior to spectroscopic analysis.

Thermal stability

The activities of PD and Δ19PD appeared to decrease exponentially with time (data not shown) and yielded half-lives at 95°C of 2 h and 55 min, respectively. Both proteins, however, retained full activity even after 20 h at 70°C. By comparison, E. coli CM-PD exhibited a half-life of 6 min at 40°C. Our values are within the range reported for other proteins isolated from thermophilic organisms (Duewel et al. 1999; Hansen et al. 1999; Iyer et al. 2002; Park et al. 2005), although results are dependent on protein concentration and buffer components.

The room temperature far-UV CD spectra for PD, Δ19PD, and CM-PD exhibited two local minima at 208 nm and 222 nm (Supplemental Fig. 2) and are typical for proteins that contain a significant content of α-helical structure. Helical contents predicted from the primary sequences of E. coli CM-PD and A. aeolicus PD (or Δ19PD), using several commercially available programs, were ∼60% and 50%, respectively. We attempted to obtain values for the apparent melting temperatures (T_m) for the proteins by measuring the ellipticity at 222 nm of the samples when heated from 25° to 95°C. However, under conditions that yielded a very cooperative and irreversible temperature-dependent unfolding curve for E. coli CM-PD (T_m = 57°C) (Supplemental Fig. 2, inset), the thermostable PDs showed a gradual but steady loss of signal, and post-transition baselines could not be established (T_m values > 95°C). Most of the CD signal (>95%) could be regained upon cooling (data not shown). However, activity measurements at 30° and 55°C on this sample showed that K_m values for prephenate were increased approximately twofold and k_cat ∼20% and implied that refolding to the native structure was not complete.

A T_m value of 108°C in the absence of denaturant was found for PD using VP-capillary differential scanning calorimetry (DSC) (Fig. 3). For comparison, an apparent T_m of 57°C was obtained for E. coli CM-PD. Similar differences in T_m values have also been reported using DSC for a hyperthermophilic archaeal acylphosphatase from Pyrococcus horikoshii versus its mesophilic counterparts (Cheung et al. 2005). For both PD and CM-PD, however, the transitions were noncooperative. A small shoulder at 95°–100°C for PD and a broad pretransition baseline for CM-PD indicated a net loss in protein hydrogen bonding, which may be attributed to subunit separation either preceding or following unfolding (Cooper 1999). Of note, the T_m obtained for CM-PD by DSC was in agreement with values of ∼56°C and ∼57°C previously determined by CD (reported here) and by variable temperature-Fourier transformed infrared (VT-FTIR) spectroscopy (R. Aponte, unpubl.), even though the experimental conditions were markedly different.

Figure 3.

Thermally induced unfolding of A. aeolicus PD and E. coli CM-PD by capillary differential scanning calorimetry.

Denaturation studies with Gdn-HCl

The stability of Δ19PD was further addressed by monitoring Gdn-HCl-induced protein unfolding using the CD signal at 222 nm as a probe of α-helical secondary structure and using tryptophan fluorescence as a probe of tertiary structure. Gdn-HCl was selected as the denaturant after noting that Δ19PD did not fully unfold in urea, even at concentrations up to 10 M.

As shown in Figure 4A, the loss of secondary structure as a function Gdn-HCl concentration yielded a multistate transition with a distinct plateau between 3 and 4 M Gdn-HCl, and a sharp decrease in fraction folded from 4 to 5 M denaturant. A value for the midpoint in the transition from folded to unfolded conformation (D_1/2) of 4.8 M was estimated from the plot. Dehydrogenase specific activity increased 220% at 0.5 M Gdn-HCl but was markedly reduced by 3 M denaturant. The presence of 0.5 M Gdn-HCl also had inverse effects on the affinity for prephenate and NAD⁺, causing a fourfold increase in K_m for prephenate (to 154 μM) and a twofold reduction for NAD⁺ (to 21 μM), relative to the values in the absence of Gdn-HCl (41 μM and 53 μM, respectively). The increase in specific activity in the presence of 0.5 M Gdn-HCl did not overcome the decrease in affinity for prephenate.

Figure 4.

The effect of Gdn-HCl on intrinsic fluorescence, CD signal, and enzyme activity of A. aeolicus Δ19PD. (A) Plot of fraction folded as a function of Gdn-HCl concentration probed by CD at 222 nm (△) and by tryptophan fluorescence emission at 317 nm (•). Fraction unfolded (F_u) was calculated as described by Pace and Scholtz (1998). Enzymatic activity in the presence of Gdn-HCl was determined at 30°C using 1 mM prephenate and 2 mM NAD⁺. The left axis represents the percent residual activity (□). (B) Selected fluorescence emission spectra of Δ19PD during Gdn-HCl-induced unfolding. Excitation was at 295 nm. Denaturant concentrations were 0 M (thick solid line), 1 M (small-dashed line), 2 M (large-dashed line), 2.9 M (dotted line), 3 M (dashed/dotted line), 4 M (large-dashed line), 4.7 M (dashed/dotted/dotted line), 5 M (thick dotted line), and 6 M (thick dotted/dashed line). The standards (6 μM NATA / 30 μM NAYA) in 6 M Gdn-HCl are shown as triangles.

Intrinsic tryptophan fluorescence is very sensitive to the hydrophobic or hydrophilic environments of a folded protein and is a good probe of changes in the accessibility of fluorophores as a function of denaturant. Δ19PD contains two tryptophan residues per monomer, at positions 190 and 259, which can contribute to the fluorescence spectrum of the protein. Excitation at either 280 (data not shown) or 295 nm (Fig. 4B) resulted in an unusual emission spectrum, with two maxima at 317 nm and 330 nm. These results suggested that the fluorescence emission is dominated by tryptophans residing in hydrophobic environments, one very hydrophobic. The bimodal emission spectra decreased and shifted to a single peak at 350 nm in the presence of 6 M Gdn-HCl as the tryptophans became solvent-exposed, and resonance energy transfer from one or more of the 10 tyrosine molecules per monomer was alleviated in the unfolded protein. A spectrum of 6 μM NATA yielded an emission maximum of 355 nm and indicated that, even at 6 M Gdn-HCl, one or more of the protein tryptophan residues were not completely solvent-exposed. Although low concentrations of Gdn-HCl had a significant effect on kinetic parameters of the reaction, inspection of the fluorescence emission spectra between 0 and 1 M Gdn-HCl indicated no shift in the emission maximum at either 317 or 330 nm. When the results from both fluorescence and CD measurements were replotted as percent folded versus Gdn-HCl concentration, the two data sets were similar but not coincident (Fig. 4A).

Gdn-HCl-induced unfolding of Δ19PD appeared reversible, as determined spectroscopically and enzymatically (data not shown). Refolded protein yielded fluorescence spectra similar if not identical to those obtained prior to unfolding, after correction for enzyme concentration differences. Moreover, >95% of the enzyme activity was recovered after refolding by dilution.

To determine if dimeric Δ19PD denatured via a pathway involving a compact folded monomer and/or a more loosely folded dimer, size exclusion chromatography was performed in the presence of 0–5 M Gdn-HCl at 25°C (data not shown). A single peak (dimeric enzyme) was observed up to 2 M Gdn-HCl, which then shifted steadily to shorter retention times at higher denaturant concentrations, indicating progressive unfolding of the dimer. By 5 M Gdn-HCl, a second peak with a long retention time (unfolded monomers) was also resolved. Above 3 M Gdn-HCl, however, a high molecular weight form of the enzyme (not in the void volume) gradually appeared. Taken together, the data suggest that the quaternary structure of Δ19PD is very stable; dimeric Δ19PD unfolds prior to subunit separation, but this unfolding accompanies the formation of an oligomeric form. All species then denature to unfolded monomers. The complexity of this pathway precludes fitting data to a model, which would yield thermodynamic parameters for the unfolding reaction.

The binding of 1-anilino-8-napthalene sulfonic acid (ANS) to Δ19PD was performed in the presence of increasing concentrations of Gdn-HCl. Results agreed with our findings above and also showed that none of the partially unfolded subunits resembled a molten globule state, which is characterized by a partial loss of tertiary structure while still retaining significant secondary structure (Ptitsyn 1995). ANS is reported to bind well to this form to yield a large increase in fluorescence intensity and a blue shift in emission maximum. A blue shift in emission maximum occurred from 0 to 2 M Gdn-HCl (Fig. 5); however, this did not coincide with a large increase in fluorescence intensity. Above 2 M Gdn-HCl fluorescence intensity decreased, and this was accompanied by a gradual red shift in emission maximum (2–8 nm) until large changes were observed by 5 and 6 M denaturant (16 and 30 nm, respectively) with protein unfolding.

Figure 5.

Emission spectra of ANS in the presence of Δ19PD at different Gdn-HCl concentrations. ANS was excited at 370 nm and fluorescence spectra were recorded from 400 to 580 nm (410–540 nm shown) as monitored in the presence of Gdn-HCl at 0 (thick solid line), 1 (dashed line), 2 (thick dashed/dotted line), 3 (dotted line), 4 (dashed/dotted line), 5 (thick dashed line), or 6 M (solid line).

Kinetic properties

Enzyme assays with purified PD and Δ19PD confirmed expectations that tyrA specifies an NAD⁺-dependent prephenate dehydrogenase. The enzymes followed Michaelis-Menten kinetics, although substrate inhibition was noted at very high concentrations of prephenate (but not NAD⁺). The kinetic parameters for the PD reaction at three different temperatures are shown in Table 2. Values of k_cat/K_m increased only ∼7- to 10-fold between 30° and 80°C; the dramatic increase in k_cat (∼30-fold) was offset by a decrease in the apparent binding affinity for substrates. For the most part, PD and Δ19PD were equally effective catalysts, as indicated by comparison of k_cat/K_m values. However, N-terminal extensions appeared to cause small but reproducible effects on the values for kinetic parameters; both the k_cat and K_m for prephenate for Δ19PD were twice that for PD, notably at 80°C. The purified enzymes did not possess CM activity when assayed at 55°C with 1 mM chorismate, nor were the activities inhibited by the mutase transition state analog, endo-oxabicyclic diacid (120 μM), or by 1 mM chorismate when assayed with 0.1 mM prephenate and 2 mM NAD⁺.

Table 2.

Steady-state kinetics parameters for PD and Δ19PD catalyzed reactions at 30°, 55°, and 80°C

Preliminary studies indicated that Δ19PD did not efficiently utilize L-arogenate (NAD⁺ as a cosubstrate), primarily due to a poor affinity for this substrate (Table 2). Values of k_cat/K_m with L-arogenate increased with temperature but were reduced ∼2.5 orders of magnitude compared with those with prephenate. E. coli CM-PD (300 μg) did not yield a measurable reaction rate at 30°C in the presence of 44 mM arogenate and 4 mM NAD⁺. This finding is in contrast to the detectable, albeit poor rates reported previously with racemic or L-arogenate (Ahmad and Jensen 1987; Turnbull et al. 1991b). In this study, negligible rates (<0.5%) were obtained for either enzyme using NADP⁺ (1 mM) with prephenate (0.5 mM) or L-arogenate (44 mM).

Both PD and Δ19PD were very sluggish enzymes at temperatures <45°C and yielded a temperature optimum of ∼95°C (PD shown in Supplemental Fig. 3). This is near the physiological optimum growth temperature of the organism (Deckert et al. 1998). An Arrhenius plot was linear from 30° to 85°C, indicating a single rate-limiting step, with an activation energy (E_a) of 61.8 kcal/mol. Both PD and Δ19PD behaved identically in this regard. Interestingly, when either enzyme was assayed using low protein concentrations (<10 μg/mL) and at lower temperatures (30° and 55°C), progress curves were punctuated by a significant lag (30 sec) before initial velocities were attained. This lag was observed whether enzyme was preincubated with either substrate or the reaction was initiated with enzyme. The lag was minimized by increasing the protein concentration or conducting assays at high temperatures and was somewhat reduced by adding 0.2 mg/mL bovine serum albumin. Interestingly, no lag was observed when L-arogenate was the substrate, although product (tyrosine) inhibition was apparent.

Initial velocity patterns constructed at 55°C with either prephenate or NAD⁺ as the variable substrate over the concentration range examined yielded plots that intersected to the left of the Y-axis (Supplemental Fig. 4). Kinetic parameters obtained from the fit of these data were in reasonable agreement with those obtained using the Michaelis-Menten equation. These results are consistent with a sequential kinetic mechanism resulting from the formation of the ternary complex before product release. Furthermore, over a concentration range of substrates that reportedly yielded concave upward kinetic plots for the PD activity of E. coli CM-PD (Turnbull et al. 1990), the initial velocity patterns for the A. aeolicus PD-catalyzed reaction were linear.

PD and Δ19PD demonstrated high specific activity within a very narrow pH range. The optimal pH was 7.5 at 55°C (data not shown) when tested between pH 5.8 and 9.3, using the 3-component buffer system of 0.05 M 2-morpholinoethanesulfonic acid (MES), 0.05 N-ethylmorpholine, and 0.1 M diethanolamine, (containing 0.15 M NaCl), plus 1 mM prephenate and 2 mM NAD⁺. The enzymes were most active when assayed in buffer containing 100–250 mM NaCl; 48% of maximal activity was observed at 2 M NaCl. The proteins precipitated slightly when stored in the absence of salt. Hence, a concentration of 75 mM NaCl or above was used in all buffers for spectroscopic and enzymatic measurements. Reaction rates of purified PD recorded at 55°C were identical whether using 50 mM potassium phosphate, 50 mM HEPES, or the three-component buffer system (all buffers contained 0.15 M NaCl). No effect on enzymatic activity was observed upon the addition of EDTA (1 mM), MgCl₂, and ZnCl₂ (25 mM) to standard reaction mixtures at 55°C. However, the enzymes were weakly inhibited by cobalt, decreasing to a limiting value of 65% of the maximal activity at 0.1 mM CoCl₂.

Effects of tyrosine

PD was inhibited up to 60% by 1 mM L-tyrosine (Supplemental Fig. 5), when assayed at 30° or 55°C in the presence of 2 mM NAD⁺ and at a prephenate concentration 4×K_m at each temperature. This value increased to 90% inhibition at 80°C, which is similar to the pattern for E. coli CM-PD at 30°C obtained using similar ratios of substrate concentration to K_m (Turnbull and Morrison 1990). These trends were also noted for Δ19PD, although the degree of inhibition was lower. When the data were plotted as specific activity as a function of tyrosine concentration, however, activity decreased to a limiting value of ∼4 U/mg (PD) or ∼26 U/mg (Δ19PD) at 55° and 80°C, respectively (data not shown). Hence, the apparent increase in inhibition at the higher temperature is only relative. Hudson et al. (1983) reported that the binding of L-tyrosine to E. coli CM-PD is enhanced by NAD⁺ and vice versa. Nevertheless, doubling the concentration of NAD⁺ to 4 mM did not result in any further inhibition of activity by tyrosine at any of the three temperatures tested (data not shown). Tyrosine inhibition was also examined with prephenate as the variable substrate at fixed concentrations of NAD⁺ of 1 mM and 2 mM (Fig. 6). Double reciprocal plots were concave upward in the presence of L-tyrosine. Deviations from linearity were more pronounced at the higher concentrations of NAD⁺ and lowest concentrations of prephenate. Similar results have also been reported for PD of E. coli CM-PD by Turnbull et al. (1991b)^, Hudson et al. (1983), and Christopherson (1985) and are consistent with tyrosine promoting cooperative interactions between the subunits. This is evidenced by a fit of the velocity data in Figure 6 (2 mM NAD⁺) to the Hill equation to yield Hill coefficients of 1.0, 1.2, and 1.8 with 0, 0.1, and 0.5 mM L-tyrosine, respectively. Lines in the double reciprocal plot did not appear to intersect on the Y-axis, indicating that, at this assay temperature, tyrosine was likely binding to an allosteric site.

Figure 6.

Double reciprocal plots of the inhibition of PD by L-tyrosine at 55°C. PD activity was assayed at tyrosine concentrations of 0 (♦), 0.1 (▪), and 0.5 mM (▴) with NAD⁺ kept at 2 mM and at 0.5 mM tyrosine with 1 mM NAD⁺ (×); 1/v is expressed as μmol NADH formed/min/mg.

To determine if inhibition by tyrosine is accompanied by tetramer formation as previously noted for CM-PD (Hudson et al. 1983), size-exclusion FPLC was performed at ambient temperature with a mobile phase containing 1 mM NAD⁺ and 1 mM L-tyrosine, and the results were compared with those in the absence of ligands. This assay was performed on E. coli CM-PD, A. aeolicus PD, and Δ19PD (50–60 μM monomer) (data not shown). The mobilities of the standards were essentially identical in the presence and the absence of ligands. Single peaks were resolved for all three proteins. CM-PD underwent a shift in retention time corresponding to a molecular weight increase from 89 kDa (no ligands) to 159 kDa (with ligands), indicating tetramer formation. In contrast, neither PD (54 kDa) nor Δ19PD (52 kDa) showed altered retention times when ligands were present (data not shown).

Active site binding studies

The folded state and the degree of accessibility of the two tryptophan residues in Δ19PD were probed further by acrylamide fluorescence quenching experiments, performed in the presence of different concentrations of Gdn-HCl. Acrylamide is a polar nonionic agent that can access both surface and buried tryptophans, except those that are deeply buried within the protein core (Eftink and Ghiron 1981). Increasing concentrations of acrylamide versus fluorescence intensity showed a progressive decrease and slight downward curvature, indicating that the two tryptophan residues are not in the same environment (data not shown). Modified Stern-Volmer plots enabled determination of the fraction of fluorescence accessible to quenching (f_a) and the quenching constant K_Q at the different concentrations of Gdn-HCl. The results (Table 3) revealed ∼50% quenching of tryptophan fluorescence by acrylamide (f_a = 0.5), consistent with the accessibility of one of the two tryptophan residues per monomer in Δ19PD. Only in the presence of 6 M Gdn-HCl does the second tryptophan become accessible. A quenching experiment was then performed in the absence of denaturant but with iodide, a large polar anion that can access surface tryptophan residues. A value of f_a of 0.05 was obtained and indicated that the tryptophan that was accessible to acrylamide in the native protein was not accessible to iodide.

Table 3.

Stern-Volmer quenching constants and percentage of tryptophan fluorescence quenched by acrylamide and KI in the presence of denaturant or substrates at 30°C

Steady-state fluorescence measurements (Fig. 7A,B) and acrylamide quenching experiments (Table 3) were performed in the presence of NAD⁺ or prephenate, in order to determine if one of more tryptophan residues were in or near the active site and, if so, to obtain a binding constant for the ligand. NAD⁺ itself is a strong quenching agent of intrinsic tryptophan fluorescence of Δ19PD, which is illustrated in Figure 7B (inset). Thus, titration of the change in fluorescence intensity as a function of NAD⁺ concentration (Fig. 7A) yielded a dissociation constant for NAD⁺ from the binary complex (K_d) of 1.42 ± 0.12 μM. In reasonable agreement, quenching of tryptophan fluorescence intensity by acrylamide in the presence of 0.5 μM (<K_d) and 5 μM (>K_d) NAD⁺ yielded values of f_a of 0.45 and ∼0, respectively. These data were consistent with bound NAD⁺ offering complete protection against acrylamide quenching. Prephenate quenched tryptophan fluorescence intensity only slightly, although the small change in intensity could be fit to a K_d of 32.6±1.94 μM (Fig. 7B and inset). Again, binding to the active site was verified by the observation that 300 μM prephenate offered significant protection against quenching of fluorescence intensity by acrylamide (f_a = 0.06).

Figure 7.

Changes in fluorescence intensity of Δ19PD upon binding NAD⁺ or prephenate. (A) NAD⁺ was varied from 0.2 to 20 μM and Δ19PD fixed at 0.24 μM monomer. Only 13 of 25 data points are plotted for clarity. (B) Prephenate was varied from 0 to 480 μM and Δ19PD fixed at 1.6 μM monomer. The intrinsic tryptophan fluorescence was observed by excitation at 295 nm and measuring the emission from 300 to 400 nm. Change of fluorescence intensities (ΔF) at 333 nm was corrected and plotted vs. concentration. The dissociation constants (K_d) for NAD⁺ and prephenate (1.42 ± 0.12 μM and 32.6 ± 1.94 μM, respectively) were determined by fitting the data to the Michaelis-Menten or the quadric equations (similar results obtained). Similar results were also obtained for the titration with NAD⁺ using 1.6 μM or 0.24 μM monomer. The inset shows the emission fluorescence spectra of Δ19PD (1.6 μM monomer) in the absence (thick line) and the presence of 300 μM prephenate (dashed/dotted line) or 20 μM NAD⁺ (dotted line) using an excitation wavelength of 295 nm. Fluorescence intensity is shown as F.I. in arbitrary units.

Discussion

The heterologously expressed TyrA protein from the hyperthermophilic bacterium A. aeolicus functions in vitro as an NAD⁺-dependent cyclohexadienyl dehydrogenase, but very poorly. Hence, its annotation as a PD in this report is to reflect the much preferred substrate, prephenate. Those TyrA proteins that effectively utilize both L-arogenate and prephenate, such as from P. stutzeri (Xie et al. 2000), Z. mobilis (Zhao et al. 1993), and P. aeruginosa (Xia and Jensen 1990), do so with values of k_cat/K_m for L-arogenate and prephenate that are within an order of magnitude of one another.

A. aeolicus PD is dimeric, which agrees with the quaternary structure established by solution studies for other purified monofunctional TyrA proteins, including arogenate dehydrogenase from Synechocystis (Bonner et al. 2004), two of the cyclohexadienyl dehydrogenases mentioned above, and the genetically engineered monofunctional PDs from E. coli (Chen et al. 2003) and E. herbicola (Xia et al. 1992a). Our findings are in keeping with the crystal structure of dimeric Δ19PD complexed with NAD⁺, which shows that the active site, formed at the interdomain cleft of the NAD⁺-binding domain and C-terminal domain, contains residues that are shared between monomers (see Fig. 5 in Sun et al. 2006). This arrangement is in contrast to that of E. coli CM-PDT for which it has been established through gel filtration experiments of independently expressed domains that dimerization is only through adjacent subunits within the mutase portion of this bifunctional enzyme (Zhang et al. 1998).

In addition to being more resistant to heat denaturation than their mesophilic counterparts, thermophilic enzymes generally exhibit low activity at ambient temperatures and higher activity at elevated temperatures (Zuber and Friedman 1978). A. aeolicus PD fulfills these criteria as its maximal activity is achieved at 95°C or higher, with a 33-fold increase in k_cat relative to that at 30°C, a half-life at 95°C of ∼2 h, and an estimated T_m value of ∼108°C. PD, to our knowledge, is the most thermal stable enzyme from A. aeolicus reported to date. In contrast, significantly lower temperature optima and only moderate half-lives at 40°C have been found for E. coli CM-PD (this work) and for other TyrA proteins (Xia and Jensen 1990; Xie et al. 2000; Bonner et al. 2004). The half-life of Δ19PD is about half that of PD, at 95°C, indicating that Δ19PD is globally more unstable or “looser.” This is surprising, as the results from the protein purification, gel filtration, and secondary structure prediction (in the present study) and those of limited proteolysis of PD yielding a crystallizable domain (Sun et al. 2006), together, are consistent with PD adopting a compact structure with a floppy N-terminal region. In further support, the kinetic parameters of PD and Δ19PD-catalyzed reactions are similar but not identical, notably at high temperatures (Table 2). The N-terminal region appears to provide some structural benefits, although its role is not apparent from the crystal structure; the N-terminal residues of Δ19PD are unordered and far removed from the active site (Sun et al. 2006).

The crystal structure of Δ19PD reveals that the dimerization interface is extensive throughout the highly helical, 100-residue C-terminal domain and is supported by hydrophobic contacts between monomers (Figs. 2 and 3 in Sun et al. 2006). Thus, it should contribute greatly to the stability of the dimer. Accordingly, Gdn-HCl-induced denaturation commences with the unfolding of the dimer as monitored by the noncoincidence of changes in tertiary and secondary structure (Fig. 4) and by gel filtration. Interestingly, E. coli CM-PD is also predicted to contain an α-helical rich dimerization domain within the C-terminal region of the protein (PredictProt) and has been reported previously to unfold significantly before dissociation into monomers (Christendat and Turnbull 1999). Results in the present study suggest that the α-helical structure of Δ19PD is very stable to Gdn-HCl denaturation; D_1/2 of 4.8 M is about twice that reported for E. coli CM-PD under similar conditions (Christendat and Turnbull 1999). However, this increased stabilization stems from, in part, retention of PD's secondary structure as an oligomeric species at the higher concentrations of denaturant. No stability studies have been reported for mesophilic monofunctional PDs. For comparison, however, a report contrasting monofunctional CMs from E. coli and from the thermophilic archaea Methanococcus jannaschii (both of which are highly α-helical) also showed that the dimeric proteins unfolded substantially before dissociation into unfolded monomers in the presence of Gdn-HCl (MacBeath et al. 1998). Thermophilic CM, with a D_1/2 of ∼4.8 M, was calculated to be more stable than the mesophilic protein by ∼5 kcal/mol, by fitting coincident CD and fluorescence data to a 2-state model.

Results from our fluorescence experiments can be interpreted in terms of the structure of Δ19PD complexed with NAD⁺ (Sun et al. 2006), although it should be considered that the conformation of the free protein may be different than in its liganded state. Trp190, on helix-7, is part of a β₇-α₇-β₈ motif that is appended to the central β-sheet of the NAD⁺-binding site, and its side chain is buried in the core of the main Rossman fold. Trp259, on helix-11 in the C-terminal domain, is one of several residues lining the wall of the prephenate binding pocket in the shared active site and is ∼16 Å from Trp190. Neither Trp259 nor Trp190 are surface-exposed within the dimer, in keeping with the substantially blue-shifted emission maxima relative to tryptophan free in solution, and their inaccessibility to quenching by KI. Of the two, Trp190 is completely buried and is not accessible to quenching by acrylamide. Furthermore, Gdn-HCl-induced oligomerization may prevent Trp190 from becoming solvent-accessible until substantial protein unfolding at 6 M denaturant (Table 3). As expected, catalytically relevant portions of prephenate and NAD⁺ are in close proximity. The side chain of Trp259 is within 5 Å of either substrate, and that of Trp190 ∼8 Å and 12 Å from NAD⁺ and prephenate, respectively. Accordingly, changes in fluorescence intensity emission occur with the binding of either NAD⁺ or prephenate. It is likely that binding either substrate can prevent quenching of Trp259 emission by acrylamide. Deciphering the contributions of individual tryptophan residues awaits analysis of site-directed variants.

The intersecting initial velocity patterns obtained at 55°C for A. aeolicus PD, and changes in tryptophan fluorescence noted at 30°C from the combination of prephenate or NAD⁺ with Δ19PD, indicated that the reaction follows a sequential kinetic mechanism with substrates adding to the enzyme in a random fashion. However, substrate binding may not be in rapid equilibrium. In such a mechanism, the Michaelis constants for substrates (K_m) are true dissociation constants of substrates (K_d) and should yield the same value. Our finding that the K_d for NAD⁺ (∼1.4 μM) determined thermodynamically is over 1 order of magnitude lower than the kinetically derived K_m value (∼25 μM) suggests that a step other than catalysis may be rate-limiting, at least at temperatures lower than optimum for the enzyme. In support of this, a significant lag in attaining linear initial velocity conditions was noted at the lower temperatures and is consistent with a slow conformational change upon the binding of substrates. Surprisingly, the Arrhenius plot showed no break at temperatures between 30° and 85°C, which would typically indicate a change in rate-limiting step (Buchanan et al. 1999). Full understanding of this kinetic mechanism awaits the results from comprehensive product and dead-end inhibition studies at temperatures close to the optimum for the enzyme. Although a rapid equilibrium random kinetic mechanism has been reported for CM-PD from E. coli (Sampathkumar and Morrison 1982) and A. aerogenes (Heyde and Morrison 1978), and arogenate dehydrogenase from A. thaliana (Rippert and Matringe 2002b), a steady-state random mechanism, but with a preferred order for arogenate binding first, has been proposed for the monofunctional arogenate dehydrogenase from Synechocystis (Bonner et al. 2004).

End-product inhibition of PD provides a major regulatory control in the pathway of tyrosine biosynthesis (Champney and Jensen 1970; Turnbull and Morrison 1990). In this report, we have shown that the enzyme is inhibited by tyrosine over a broad temperature range and yields kinetics in the presence of tyrosine that are suggestive of positive cooperativity between subunits at 55°C. However, inhibition does not appear to accompany a shift in equilibrium from active dimer to inactive tetramer promoted by NAD⁺ and/or tyrosine, as noted for E. coli CM-PD (Hudson et al. 1983). Our results are consistent with suggestions that this cooperativity is manifested through tertiary structural changes (Christopherson 1985; Turnbull et al. 1991b) and that tetramerization of the E. coli enzyme is an artifact observed at high protein concentrations. It is worth noting that TyrA from P. stutzeri forms tetramers at high protein concentrations but in the absence of ligands (Xie et al. 2000). While other studies report inhibition of the activity of TyrA proteins by tyrosine (Champney and Jensen 1970; Catcheside 1979; Xia et al. 1992a; Xie et al. 2000; Rippert and Matringe 2002b), this is the first investigation of ligand-induced quaternary structural changes for a monofunctional TyrA protein under conditions that should promote tetramer formation (high concentrations of protein, tyrosine and NAD⁺). Studies on CM-PDT by Ganem and colleagues (Zhang et al. 1998, 2000) also showed that this bifunctional enzyme undergoes oligomerization upon phenylalanine binding. Moreover, fusion of the N-terminal CM domain directly to the C-terminal Phe-binding domain resulted in activation of the enzyme by the end product but no oligomerization. They proposed that residues involved in the oligomerization are found within the PDT domain. Interestingly, Sun et al. (2006) reported that Δ19PD complexed with NAD⁺ crystallized as a tetramer, but only three pairs of interdomain interactions were observed. Hence, it remains to be determined if this tetrameric species is biologically relevant. The possibility arises that A. aeolicus PD does form tetramers but only at temperatures nearer to the activity optimum. At 55°C, we did not observe cooperativity in the binding of prephenate to the enzyme, although it is greatly enhanced upon tyrosine addition. It may be that, at a temperature considerably below the optimum, the protein is less flexible and thus less sensitive to changes that promote interactions between subunits.

It has been hypothesized that the lower catalytic rate of thermostable enzymes at suboptimal temperatures is due to insufficient flexibility in their active sites (Cheung et al. 2005). Addition of low concentrations of Gdn-HCl to Δ19PD resulted in a two- and fourfold increase in k_cat and K_m, respectively, for prephenate, perhaps due to increased flexibility at the active site. Similar results have been reported for other thermophilic enzymes upon the addition of low concentrations of urea or Gdn-HCl (Natrayanasami et al. 1997; Zhang et al. 1997; Kujo and Oshima 1998; Thomas and Scopes 1998; Inui et al. 1999; Zoldác et al. 2003). The same mechanism likely accounts for the small but reproducible differences in kinetic parameters between PD and Δ19PD. Moreover, changes in protein conformation within the hydrophobic NAD⁺-binding pocket of PD, induced by low denaturant concentrations, can be illustrated by the coincidence of a decrease in K_m for NAD⁺ with an increase in binding of the amphiphilic dye ANS (Fig. 5); it has been reported previously that ANS has affinity for such nucleotide-binding sites (McLaughlin 1974; Cox et al. 2003).

Analysis of many thermophilic proteins and their corresponding mesophilic homologs revealed several differences, which include a larger proportion of charged versus polar amino acids (Vogt et al. 1997; Cambillau and Claverie 2000; Szilagyi and Zavodszky 2000). A. aeolicus PD houses a significantly higher proportion of lysine and glutamate residues (10.6% and 8%, respectively) than found in the PD portion of E. coli CM-PD (3.7% and 5.9%, respectively, for example). In fact, the crystal structure of Δ19PD complexed with NAD⁺ (Sun et al. 2006) reveals a large ionic network formed by Glu275, Glu278, and Lys285 from one subunit and the same residues from the other subunit, a structural feature that Sun et al. (2006) speculated may contribute to the thermostability of the enzyme. Other parameters noted with elevated thermostability and thermal activity of proteins are an increased number of strategically located proline residues, as seen in CM from T. thermophilus (Helmstaedt et al. 2004), or a number of leucine-to-isoleucine substitutions, noted for M. jannaschii CM (MacBeath et al. 1998). However, sequence analysis of A. aeolicus PD does not appear to support these strategies. Interestingly, our sedimentation velocity data from AUC yield a low frictional coefficient of 1.22 for A. aeolicus PD suggestive of a compact protein core. Moreover, VT-FTIR results (R. Aponte, unpubl.) support this observation in showing slow H/D exchange at high temperatures, which would have likely continued well past 95°C except for instrument limitations. It would be interesting to speculate that A. aeolicus PD might have adopted elements of “sequenced-based” (ion pairs, H-bonding) and “structure-based” (packing) mechanisms for its thermostabilization of PD (Berezovsky and Shakhnovich 2005).

Interestingly, no genes encoding CM or CM-PD were identified in A. aeolicus, although pheA encoding CM-PDT was identified (Deckert et al. 1998). Pairwise sequence alignment between A. aeolicus PD and the PD domain of E. coli CM-PD showed only 18.5% sequence identity. However, functional studies on CM-PD indicated that residues deemed catalytically important for dehydrogenase activity were present in A. aeolicus PD (see Supplemental Fig. 1; Sun et al. 2006). These features are also well documented in the TyrA protein family by Jensen and colleagues (Song et al. 2005). Ultimately, A. aeolicus PD was selected for study because the biophysical characteristics associated with thermophilic proteins potentially aid in crystallization. We believe that the structure of this thermophilic monofunctional PD could provide a basis for modeling mesophilic relatives within the TyrA enzyme family.

Materials and methods

Materials

Thrombin was obtained from the Ontario Cancer Institute, University of Toronto. Prephenate (sodium salt), chorismate (free acid), and arogenate (barium salt) were prepared as previously described (Dudzinski and Morrison 1976; Bonner et al. 1990; Rieger and Turnbull 1996), while NAD⁺ and NADP⁺ (free acid) were obtained from Roche. High purity of the substrates was confirmed by either mass spectrometry or NMR. Concentrations of stock substrate solutions were determined using published extinction coefficients (Dawson et al. 1986) and/or enzymatic end-point analysis. N-Acetyl-L-tryptophanamide (NATA) and N-acetyl-L-tyrosinamide (NAYA) were purchased from Sigma, and concentrations of stock solutions were determined spectrophotometrically (Edelhoch 1967). All other chemical reagents were obtained commercially and were of the highest quality available. Reagents and enzymes for molecular biology were provided by Stratagene, New England Biolabs, and Promega. E. coli strains DH5α (Life Technologies Inc.) and BL21(DE3) Gold (Strategene) were used for plasmid production and for protein expression, respectively. The gene encoding A. aeolicus PD was cloned into the E. coli expression vector pET-15b (Novagen). Dr. A. Edwards at the Ontario Cancer Institute kindly donated the helper plasmid pMagik, while A. aeolicus genomic DNA was generously provided by Drs. Robert Huber and Karl O. Stetter from the Universität Regensburg, Germany. Oligonucleotides were obtained from BioCorp Inc. at standard purity. All relevant portions of constructed plasmids were confirmed by DNA sequencing through the facilities at Bio S & T.

Construction of plasmids

All nucleic acid manipulations were performed using standard methods (Sambrook and Russell 2001) or as suggested by the manufacturer. Primers P1 (5′-GCGGCGGCCCATATGGCTATCCTCTCCAGTAT-3′) and P2 (5′-GCGCGGATCCTCAATCTATCTCCATTCTCTTTAA-3′) corresponded to the 5′ and 3′ ends of the tyrA gene (AAC07589 from GenBank), respectively, and incorporated flanking restriction sites NdeI (for P1) and BamHI (for P2). The putative tyrA gene was amplified by PCR from A. aeolicus genomic DNA, and the PCR product (971 bp) was gel-purified and treated with NdeI and BamHI. The restricted fragment was isolated from 1% agarose utilizing the QIAquick Gel Extraction Kit (Qiagen) and then ligated into pET-15b and transformed into competent E. coli DH5α. Transformants were selected on Luria-Bertani (LB)/agar plates containing 100 μg/mL ampicillin, and plasmid DNA was isolated and subjected to DNA sequencing. One clone with the correct sequence of tyrA (pRA-PD-3) was used to transform E. coli BL21(DE3) Gold cells. The construction of Δ19PD is described elsewhere (Sun et al. 2006).

Production and purification of proteins

Recombinant E. coli CM-PD (∼30 U/mg) was expressed and purified as described elsewhere (Christendat and Turnbull 1999). E. coli BL21(DE3) Gold cells harboring pMagik and pRA-PD-3 were grown in 50 mL of LB medium supplemented with 100 μg/mL ampicillin and 50 μg/mL kanamycin at 30°C for 15 h with shaking, and diluted into 1.5 L of the same medium. After growth to an OD₆₀₀ of 0.6, 0.4 mM 1-thio-β-D-galactopyranoside was added. The cells were incubated further for an additional 5 h at room temperature and then overnight at 18°C. Cells were harvested by centrifugation and resuspended in 15 mL/L culture of ice-cold buffer A (50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [HEPES], 0.5 M NaCl, 5% glycerol at pH 7.5) supplemented with 5 mM imidazole, Complete (Roche protease inhibitor cocktail, one tablet per 50 mL suspension), 1 mM benzamidine, and 0.5 mM phenylmethyl sulfonyl fluoride. The cells were disrupted by two passages through a Thermo Spectronic French Press at 18,000 psi, with additional benzamidine added after the first passage. Insoluble material was removed by centrifugation at 100,000g for 40 min at 4°C. The cell-free extract was incubated in an 85°C water bath for 10 min and centrifuged again, and the supernatant was applied to a 50 mL Q-Sepharose anion exchange column (Amersham) equilibrated with buffer A containing 5 mM imidazole. All chromatography was performed at room temperature. The Q-Sepharose column was washed with buffer A containing 30 mM imidazole, and the eluate was loaded onto a 15 mL Superflow Ni-NTA column (Qiagen) at a flow rate of 1 mL/min. Nickel resin was washed extensively with the above buffer, and bound protein was eluted with buffer A containing 300 mM imidazole. Fractions were supplemented with 1 mM EDTA and 0.5 mM dithiothreitol (DTT). Those containing PD activity were pooled, and thrombin was added at a final protein:thrombin ratio of 1000:1 (w/w). Following dialysis (Spectrapore, 12 K cutoff) for 2 h at room temperature, then overnight at 4°C against buffer A with 0.5 mM Tris(2-carboxyethyl) phosphine hydrochloride (TCEP-HCl), the sample was reapplied onto the Ni-NTA column. Unbound PD was rechromatographed on a 1 mL HiTrap benzamidine FF column (Amersham Bioscience), concentrated to 2–10 mg/mL (Amicon Ultra-15), and stored at −86°C in buffer A containing 5 mM DTT (storage buffer) and 1 mM benzamidine. PD activity was detected in fractions by enzyme assay and SDS-15% PAGE (Laemmli 1970) with Coomassie Blue staining. His-tagged protein was detected by treating selected denaturing gels with Pro-Q Sapphire 365 Oligohistidine Gel Stain (Molecular Probes) according to the manufacturer's instructions. Δ19PD was purified and stored as described for PD, except thiol reducing agents were omitted in the purification procedure. PD and Δ19PD were further purified by size exclusion chromatography (outlined below) prior to spectroscopic studies.

Determination of enzyme activity and protein concentration

The oxidative decarboxylation of prephenate or arogenate in the presence of NAD⁺ was followed at 340 nm, while the conversion of chorismate to prephenate was monitored at 274 nm (Turnbull et al. 1990). The reactions (total volume 1.0 mL) were monitored continuously by using either a GBC UV/VIS model 918 or a Varian Cary 50 spectrophotometer, both equipped with a thermostated cuvette holder.

Standard activity assay for A. aeolicus PD and Δ19PD was measured in a reaction buffer of 50 mM HEPES, 150 mM NaCl (pH 7.5), at saturating concentrations of NAD⁺ (2 mM) and prephenate (1 mM). For assays at 30°C and 55°C, buffer was incubated at the fixed temperature (2 min) and followed by the addition of an appropriate amount of enzyme (2 min) and NAD⁺ (30 sec), and then the reaction was initiated with prephenate. For assays performed between 80°C and 95°C, the incubation times for the enzyme and NAD⁺ were reduced to 1 min and 15 sec, respectively. Reactions with L-arogenate were initiated with enzyme after preincubation of reaction buffer and substrates at the desired temperature for 2 min. At each step, components were gently mixed by inversion of the cuvette. The pH of the buffers was checked at the working temperature. All substrates were at room temperature prior to their addition. E. coli CM-PD activity was measured at 30°C as described previously (Turnbull et al. 1990).

Enzyme half-life was determined by incubating enzyme (1 mg/mL in assay buffer) in capped Eppendorf tubes at 95°C and 70°C (A. aeolicus PD and Δ19PD) or at 40°C (E. coli CM-PD). For CM-PD, buffer was supplemented with 25% glycerol (v/v). Samples were removed at different time intervals, cooled on ice, and centrifuged for 5 min, and residual activity of the supernatant was determined by the standard assay at 55°C (PD) or 30°C (CM-PD). Protein concentration was determined after centrifugation in order to calculate specific activities.

Reaction rates were calculated from the linear portion of progress curves using the software supplied by the spectrophotometer. Values of steady-state kinetic parameters k_cat and K_m were obtained by fitting initial velocity data to the appropriate rate equations using nonlinear least-squares analysis provided by Grafit Software version 5.0 (Erathicus Software) or the programs of Cleland (1979). Substrate saturation curves were fitted using the Michaelis-Menten equation. When appropriate, the Hill coefficient was obtained from the slope of a plot of log[v/(V_max – v)] versus log [substrate], using values of v (initial velocity) between 10% and 90% V_max (maximum velocity). The linearity of the double reciprocal plots was assessed with Excel. Those that were concave upward were fitted to a parabola.

A unit of enzyme is defined as the amount of enzyme required to produce 1 μmol of product per min at the specified temperature. Turnover numbers were calculated using subunit molecular weights (kDa) of 37.01 (full-length PD), 35.13 (thrombin-cleaved PD), 36.88 (heterodimeric PD), and 33.19 (Δ19PD) assuming one active site per monomer. Protein concentration was estimated using the Bio-Rad Protein Assay Kit (Bio-Rad Laboratories) with bovine serum albumin (Sigma) as a standard.

Molecular weight determination

Mass spectrometry

Subunit molecular weight of PD and Δ19PD was determined by electrospray ionizing mass spectrometry (ESI-MS). Analysis was carried out on a Waters Micromass Q-ToF-2 mass spectrometer operating in positive-ion mode following direct infusion of samples into the Z-spray ion source. Instrument parameters were as follows: source block temperature, 80°C; capillary voltage, 3.6 kV; cone voltage, 45 V; ToF, 9.1 kV; MC, 2.1 kV. Sample preparation was adapted from a protocol for the analysis of membrane-associated proteins (Weinglass et al. 2003). Briefly, enzyme (∼100 μg) in storage buffer was resuspended in a mixture of methanol/chloroform/water (3:1:2), the sample was centrifuged 2 min at 14,000g, and the protein precipitate was washed with two volumes of methanol. The enzyme was resuspended in a solution of 30% methanol/0.2% formic acid (v/v) immediately prior to injection. Data analysis and deconvolution were performed using MassLynx 4.0 software (Waters Micromass). Calibration of the instrument was checked with [Glu]-fibrinopeptide B (Sigma). Mass shifts of ±2 mass units for PD are within the expected experimental error.

Analytical ultracentrifugation

Native molecular weight and shape of PD and Δ19PD were determined by sedimentation velocity experiments performed at 30°C in a Beckman XL-I analytical ultracentrifuge and an An60Ti rotor using absorbance detection. Purified thrombin-cleaved enzymes were exchanged into a buffer containing 50 mM potassium phosphate buffer, 0.3 M NaCl and 0.5 mM TCEP (pH 7.5) using a NAP-5 size exclusion column (Amersham), diluted in the same buffer to give a final OD₂₈₀ of 0.65 (PD) and 0.97 (Δ19PD), and loaded into 1.2 cm path-length doublesector charcoal-filled epon centerpieces. Samples were spun at 35,000 rpm and 30°C for 10 h. Absorbance (280 nm) was collected in continuous mode with a step size of 0.005 cm and five replicate readings at each point. Values for the sedimentation coefficient (s) and an average molar mass were calculated from the velocity and shape of the sedimenting boundary by fitting the time-dependent concentration profiles calculated with the Lamm equation (Lamm 1929) to the measured data. Calculations were performed on 200 scans for each protein using the program Sedfit (http://www.analyticalultracentrifugation.com/default.htm). The program Sednerp (Laue et al. 1992) was used to calculate buffer density and the protein partial specific volume (1.0153 and 0.74, respectively, at 19°C, neglecting contributions due to TCEP) and to convert s to S_20,w.

Analytical size exclusion chromatography

The native molecular weights of A. aeolicus PD, Δ19PD, and E. coli CM-PD were determined in the absence and the presence of 1 mM each of NAD⁺ and L-tyrosine at ambient temperature by a Pharmacia Akta FPLC system fitted with a Superdex G-120 column (HR 10/30, Pharmacia). Chromatography was performed with mobile phases containing 50 mM potassium phosphate, 150 mM NaCl (pH 7.5) with and without ligands at a flow rate of 0.75 mL/min and injection volume of 500 μL. Elution was monitored at 256, 280, and 290 nm, and fraction (1 mL) was assayed for enzyme activity and inhibition by tyrosine. Bio-Rad gel filtration protein standards included vitamin B₁₂ (1.35 kDa), equine myoglobin (17 kDa), chicken ovalbumin (44 kDa), bovine γ-globulin (158 kDa), and thyroglobulin (670 kDa). Void volume and total bed volume were evaluated with Blue Dextran and DTT, respectively. When chromatography was conducted in the presence of Gdn-HCl-containing buffers, PD or protein standards (0.1 mg/mL) were incubated for 20 h at ambient temperature in a buffer of 50 mM potassium phosphate, 150 mM NaCl and various Gdn-HCl concentrations (pH 7.5), and then isocratically separated in the same buffer at a flow rate of 0.4 mL/min.

Denaturation studies

Circular dichroism

Far-UV CD spectra of PD and CM-PD (21 μM and 18 μM monomer, respectively) were recorded on a Jasco-710 spectropolarimeter in either a 0.05-cm or 0.1-cm path-length circular cell connected to a thermostated circulating water bath. Protein was exchanged into 50 mM potassium phosphate, 75 mM NaCl (pH 7.5) (PPS buffer), using a NAP-5 column and then diluted to the appropriate concentration in the same buffer. For studies with CM-PD, 25% glycerol (v/v) was added to the buffer. Spectra were recorded at 25°C by averaging 10 wavelength scans from 260 to 200 nm (1 nm bandwidth) in 0.2-nm steps at a rate of 50 nm/min, and 0.25 sec response. The ellipticity at 222 nm (1 nm bandwidth) was measured from 25° to 95°C by using the instrument software controlled temperature ramping program and the following parameters: ΔT of 20°C/h, 0.2°C step resolution, and 1 sec response.

Equilibrium denaturation of Δ19PD induced by Gdn-HCl was followed by measuring the ellipticity of the sample at 222 nm in a 0.1-cm path-length cell at 30°C. Samples at each Gdn-HCl concentration were obtained by mixing PPS and 8 M Gdn-HCl in PPS (pH 7.5) in the appropriate ratio and adding enzyme to 2.8 μM. Samples (1 mL) were equilibrated at ambient temperature for 20 h in capped Eppendorf tubes, and then ellipticities were measured from 210 to 230 nm using the instrument parameters listed above. The values at 222 nm (average of five readings) were corrected for background signal from the buffer. The accurate concentration of 8 M Gdn-HCl in PPS was calculated from its refractive index (Pace and Scholtz 1998).

Steady-state fluorescence

Gdn-HCl-induced unfolding of Δ19PD was followed by fluorescence at 30°C using an Aminco Bowman Series 2 Luminescence Spectrometer equipped with a temperature-controlled cell holder. Excitation wavelengths were set to 280 nm or 295 nm, and emission scans were recorded from 300 to 400 nm. Excitation and emission slits were set to 4 nm. Measurements were performed in PPS buffer (reaction volume 2 mL) using a 1 cm × 1 cm cuvette. Incubation of Δ19PD (3 μM monomer) in Gdn-HCl-containing PPS were performed as outlined for CD experiments. Fluorescence intensities were compared with that of a solution containing NATA (6 μM) and NAYA (30 μM) having the same concentration of tryptophan and tyrosine as the protein solution. Emission spectra were corrected for buffer blank and for the inner filter effect (Lakowicz 1999). To determine if denaturation was reversible, 30 μM Δ19PD in 0 M, 3 M, and 5 M Gdn-HCl were incubated for 20 h at room temperature. An aliquot of the denatured sample was then diluted 10-fold then incubated at room temperature for 2 h. Fluorescence spectra and enzyme activities were recorded for each sample and compared with a protein incubated for the same time in the absence of denaturant. Measurements at 20, 24, and 36 h gave the same values.

Differential scanning calorimetry

Measurements were performed on a Microcal VP-capillary DSC instrument at a constant heating rate of 200°C/h. Aliquots of A. aeolicus PD and E. coli CM-PD were buffer-exchanged into 100 mM potassium phosphate (pH 7.5), 0.3 M NaCl, 5% glycerol, and 0.5 mM TCEP. Data acquisition was conducted at constant pressure (30 psi), using medium feedback with a 2-sec filter and 5 min equilibration time. Protein concentrations used were 6.0 mg/mL for PD and 10.0 mg/mL for CM-PD, in a total cell volume of 0.11 mL. Apparent melting temperatures were determined by the midpoint of thermal transition (Origin software). Rescanning of the samples indicated that the thermal denaturation of both proteins was irreversible under these conditions.

Fluorescence quenching

The titration of Δ19PD with acrylamide and KI were performed at 30°C in PPS buffer. Excitation and emission wavelengths were set at 295 nm and 340, respectively, with bandwidths of 4 nm each. Sodium thiosulfate (100 μM) was added to solutions of KI to prevent I₃⁻ formation, which interferes with tryptophan fluorescence. Defined amounts of quencher (2–5 μL) were added from a stock of 5.0 M to a 2 mL (3 μM monomer) protein solution. Acrylamide and KI did not significantly inhibit enzyme activity (<5% loss of activity) at the amounts used in the quenching experiments. To determine if substrates protected against quenching, 3 μM monomer was incubated either with NAD⁺ or prephenate (also in PPS) prior to the addition of acrylamide or KI. The effect of Gdn-HCl on fluorescence quenching was examined at 30°C by the addition of acrylamide to a reaction mixture containing 3 μM monomer and 0–6 M Gdn-HCl in PPS. Samples were incubated for 20 h at room temperature prior to measurements. Emission spectra represented the average of three scans (2 nm/sec), and were corrected for background and protein dilution and, when appropriate, the inner filter effect (Lakowicz 1999).

Fluorescence quenching data were analyzed by the Stern-Volmer equation where static quenching is neglected (Lehrer 1971): F_o/F = 1 + K_SV[Q]. F_o and F are the fluorescence intensities in the absence and the presence of the quencher, respectively; [Q] is the concentration of the quencher; and K_SV is the collisional Stern-Volner constant, which is a direct measure of the quenching efficiency. In a protein containing multiple tryptophan residues, the presence of different classes of tryptophan residues is reflected by a downward curvature in the Stern-Volmer plot. The fraction of total fluorophore accessible to the quencher in a heterogeneous system was determined using a modified Stern-Volmer plot (Lehrer 1971): F_o/(F_o − F) = 1/f_a + 1/[Q]K_Qf_a. K_Q is the modified quenching constant and f_a is the fraction of the initial fluorescence accessible to the quencher. Values for K_Q and f_a can be determined from a plot of F_o/(F_o − F) versus 1/[Q].

ANS fluorescence experiments

Δ19PD (3 μM monomer) prepared in different concentrations of PPS-buffered Gdn-HCl (as described above) were incubated with 30 μM 1-anilino-8-naphthalene sulfonic acid (ANS) in the dark at 30°C. The total reaction volume was 2 mL. Fluorescence emission spectra were recorded from 400 to 600 nm with excitation at 370 nm and using bandwidths of 4 nm. The net fluorescence enhancement due to ANS binding to the protein was obtained by subtracting appropriate blank spectra of ANS in the corresponding denaturation buffer and corrected for inner filter effects as described above. ANS did not alter enzyme activity.

Determination of dissociation constants for substrates

Values for the dissociation of NAD⁺ or prephenate from the complex with Δ19PD were determined at 30°C by monitoring the quenching of protein intrinsic fluorescence. Excitation and emission wavelengths were set a 295 nm and 333 nm, respectively, with bandwidths of 4 nm. Enzyme, NAD⁺, and prephenate were prepared in PPS buffer. Titrations were performed by the progressive addition of NAD⁺ (0.2–200 μM) or prephenate (2–480 μM) to PPS (2 mL) containing 0.24 μM or 1.6 μM monomer. The reaction was allowed to equilibrate for 5 min prior to recording measurements. The fluorescence data were corrected for inner filter effects, as well as for dilution and background fluorescence. A dissociation constant was determined by fitting the data to the Michaelis-Menten equation or the quadratic equation (shown below) (Engel 1996) using Grafit 5.0: ΔF = ΔF_m(([L_t] + [E_t] + K_d) − (([L]_t + [E_t] + K_d)² − 4[L_t][E_t])^0.5)/(2[E_t]). ΔF is the difference in fluorescence intensities in presence and absence of the titrant, ΔF_m is the maximum change in fluorescence intensity, [L_t] is the total concentration of titrant, [E_t] is the total enzyme concentration, and K_d is the dissociation constant.