The interaction of ammonia and xenon with the imidazole glycerol phosphate synthase from Thermotoga maritima as detected by NMR spectroscopy

Abstract

The imidazole glycerol phosphate (ImGP) synthase from the hyperthermophilic bacterium Thermotoga maritima is a 1:1 complex of the glutaminase subunit HisH and the cyclase subunit HisF. It has been proposed that ammonia generated by HisH is transported through a channel to the active site of HisF, which generates intermediates of histidine (ImGP) and de novo biosynthesis of 5-aminoimidazole-4-carboxamideribotide. Solution NMR spectroscopy of ammonium chloride-titrated samples was used to study the interaction of NH₃ with amino acids inside this channel. Although numerous residues showed ¹⁵N chemical shift changes, most of these changes were caused by nonspecific ionic strength effects. However, several interactions appeared to be specific. Remarkably, the amino acid residue Thr 78—which is located in the central channel—shows a large chemical shift change upon titration with ammonium chloride. This result and the reduced catalytic activity of the Thr78Met mutant indicate a special role of this residue in ammonia channeling. To detect and further characterize internal cavities in HisF, which might for example contribute to ammonia channeling, the interaction of HisF with the noble gas xenon was analyzed by solution NMR spectroscopy using ¹H-¹⁵N HSQC experiments. The results indicate that HisF contains three distinct internal cavities, which could be identified by xenon-induced chemical shift changes of the neighboring amino acid residues. Two of these cavities are located at the active site at opposite ends of the substrate N′-[(5′-phosphoribulosyl)formimino]-5-aminoimidazole-4-carboxamide-ribonucleotide (PRFAR) binding groove. The third cavity is located in the interior of the central β-barrel of HisF and overlaps with the putative ammonia transport channel.

Introduction

The imidazole glycerol phosphate synthase (ImGP-S) is a central enzyme at the branching point of two anabolic pathways and belongs to the large family of glutamine amidotransferases.1 It catalyzes one of the 11 steps in histidine biosynthesis by forming ImGP. Furthermore, it generates 5-aminoimidazole-4-carboxamideribotide (AICAR), an intermediate in the de novo synthesis of purines.2, 3 ImGP–S from bacteria consists of two separate enzymes, HisH and HisF. HisH exhibits glutaminase activity and HisF catalyzes the aforementioned cyclase reaction resulting in the formation of ImGP and AICAR (Fig. 1). For in vitro studies, ammonia produced by HisH can be substituted by ammonium salts.2, 4

Figure 1. Reactions catalyzed by the glutaminase HisH and the synthase HisF of imidazole glycerol phosphate (ImGP) synthase. The utilization of nitrogen is shown in gray. PRFAR: N′-[(5′-phosphoribulosyl)formimino]-5-aminoimidazole-4-carboxamide-ribonucleotide; AICAR: 5-aminoimidazole-4-carboxamideribotide.

The X-ray structure of ImGP-S from the hyperthermophilic bacterium Thermotoga maritima shows that HisF adopts a (βα)₈-barrel fold (see Fig. 2) with the C-terminal (catalytic) face forming the active site and the N-terminal (stability) face interacting with HisH.5, 6 The active sites of HisH and HisF are separated by a distance of ∼25 Å.6, 7 Solution NMR spectroscopy was used to assign the ¹H, ¹⁵N, and ¹³C resonances of HisF and to study the influence of ImGP binding and complex formation between the two subunits on the protein dynamics of HisF.8, 9

Figure 2. Crystal structure of HisF according to Lang et al.5 shown as ribbon plots of the molecule (top: side view, bottom: view from the activity surface into the central channel).

It has been proposed that NH₃ is transported between the active sites of HisH and HisF through an ammonia channel, which includes the interior of the central barrel of HisF.6, 7 The existence of similar channels, which protect the NH₃ molecule from the solvent and thus maintain its nucleophilic reactivity, has been postulated also for other glutamine amidotransferases.10 However, little direct experimental evidence has been provided yet for the function of such channels. Therefore, we have analyzed the interaction of ammonia and the hydrophobic noble gas xenon with HisF by means of solution NMR spectroscopy within this work. Xenon has become a well-established probe for the study of cavities within the interior of proteins (see, e.g., Refs. 11–17). Advantageously, the inert character of xenon only minimally changes the properties of the sample solution.

Results

Interaction of NH₃ with HisF and steady-state enzyme kinetics

The central channel within the crystal structure of HisF has been proposed previously to be part of the ammonia transfer route in the homologous ImGP-S from yeast.7 In particular, the highly conserved amino acid residues Thr78, Ser101, and Ser201 were suggested to function as ammonia relay stations during the channeling process. To further substantiate this hypothesis, we have studied the influence of ammonia upon the solution NMR spectra of HisF by adding NH₄Cl to the sample solution. In aqueous solutions, NH and NH₃ are forming a chemical equilibrium. We have evaluated the changes induced by increasing concentrations of NH₄Cl (0, 10, 50, 200, 400, and 578 mM) upon the assigned ¹H-¹⁵N HSQC spectrum of HisF. The NMR signals of numerous amino acids are influenced by NH₄Cl. In particular, pronounced ¹⁵N chemical shift changes are observed (see Fig. 3), which can be described by a one site binding model (see Fig. 4). A binding event would indeed result in such hyperbolic saturation curves. However, the chemical shift changes caused by NH₄Cl are not only caused by specific ammonia binding but also by the influence of the changing ionic strength of the buffer. To discriminate between these two effects, samples of equal ionic strength but different buffer composition were compared: One sample contained 200 mM NH₄Cl plus 200 mM NaCl and the other sample was prepared using 400 mM NH₄Cl. Chemical shift differences between these two samples of equal ionic strength can be ascribed to the specific effect of NH₃/NH (see Fig. 3). For most amino acid residues, the replacement of NH by Na⁺ did not influence the ¹⁵N chemical shifts as is demonstrated in Fig. 3 (top) for Cys9. Obviously, the corresponding chemical shift changes are caused solely by the ionic strength of the solution. The replacement of 200 mM NH by 200 mM Na⁺ caused an average ¹⁵N chemical shift change of only 0.02 ppm for the entire protein backbone. This is even below the experimental error of 0.035 ppm for one single measurement. Significant ammonia-induced effects, that is, values exceeding at least two times this experimental error, occurred for only six amino acid residues, namely Ile42, Ala70, Glu71, Thr78, Thr178, and Leu241. Three of the ammonia-sensitive amino acid residues are hydrophobic (Leu241, Ile42, and Ala70) and two are uncharged hydrophilic (Thr78 and Thr178). Only one is charged (Glu71), which may preferentially interact with NH. For the uncharged amino acids, however, the observed effects cannot be caused by ionic effects although it is impossible to experimentally discriminate between the influence of NH₃ and NH. Apart from Thr78, all ammonia-sensitive amino acid residues are located at the external surface of the protein, that is, they are freely accessible for ammonia. Thr78 is the only ammonia-sensitive amino acid located inside the central channel (see Figs. 2–4). The fact that Thr78 is significantly influenced by ammonia as well as its location in the central channel indicates that this amino acid residue is involved in ammonia transport, in line with previous assumptions (see above). To verify this hypothesis, we have generated the HisF-Thr78Met mutant to block ammonia flux through the channel by introducing a bulky side chain. We have then measured the catalytic activity of the mutant HisH:HisF complex, either in the presence of glutamine (cf. Fig. 1) or exogenous ammonia being provided as ammonium salt. In the latter case, HisF activity is not necessarily dependent on the ammonia channeling system. Steady-state enzyme kinetics at 25°C showed that the turnover number (k_cat) of the ammonia-dependent reaction (0.73/s) is four-fold higher than the k_cat of the glutamine-dependent reaction (0.18/s). Comparable measurements with wild-type HisH:HisF yielded no significant difference between the turnover numbers of the ammonia- and the glutamine-dependent reactions (1.2/s and 1.1/s, respectively). These results suggest that the replacement of the hydrophilic side chain of Thr78 by the voluminous hydrophobic side chain of Met around cavity III indeed impedes the transport of ammonia between the active sites of HisH and HisF. The Michaelis constants for PRFAR are similar for the ammonia- and glutamine-dependent reactions, both for the mutant (K = 5.9 μM and 2.8 μM) and the wild-type enzyme complex (K = 3.6 μM and 2.0 μM), which shows that the exchange Thr78Met does not impair substrate binding to the active site of HisF.

Figure 3. Two selected signals from the ¹H-¹⁵N HSQC spectra recorded for the salt free sample (black), in presence of 200 mM NH₄Cl plus 200 mM NaCl (gray), and in presence of 400 mM NH₄Cl (dashed black). Thr78 (bottom) differentiates between NH₄Cl and NaCl while most of the other signals exhibit purely ionic strength-induced shifts as for example found for Cys9 (top).

Figure 4. ¹⁵N chemical shift of Thr78 measured at increasing NH₄Cl concentration (rectangles). One datapoint (circle) was recorded with a buffer containing 200 mM NH₄Cl and 200 mM NaCl to differentiate between the influence of ammonia and purely ion strength-induced effects.

Specific xenon binding to HisF detected by ¹²⁹Xe NMR

Xenon is capable of interacting with proteins without changing the ionic strength of the solutions in contrast to NH₄Cl. ¹²⁹Xe NMR spectroscopy allows the distinction between specific and nonspecific xenon binding to proteins. The ¹²⁹Xe chemical shift usually exhibits a linear dependence on the protein concentration.11–14 The comparison of the slopes of these curves as observed for native and unfolded protein allows one to distinguish between specific and nonspecific xenon binding.13 If present, specific binding sites in the protein structure will be destroyed by protein unfolding. Consequently, the slope of the ¹²⁹Xe chemical shift versus protein concentration curve in the native state is then higher than in the denatured protein. In the absence of specific binding sites, protein unfolding results in an increased solvent-exposed surface, which favors nonspecific xenon binding. Therefore, protein unfolding will lead to an increasing slope of the ¹²⁹Xe chemical shift versus protein concentration curve for nonspecific binding.13 Our experiments with HisF revealed a decreasing slope of the ¹²⁹Xe chemical shift versus protein concentration curve upon unfolding (Fig. 5 and Table I). This behavior indicates specific xenon binding by HisF in its folded state. Binding of saturating concentrations of ImGP to folded HisF did not influence the slope significantly, which shows that the xenon-binding site(s) in HisF are not influenced by the presence of this reaction product (Fig. 5 and Table I).

Figure 5. ¹²⁹Xe chemical shift dependence on the concentration of HisF in the folded state (rectangles), the unfolded state (triangles), and the ligand (ImGP) bound state (open circles). The slopes of the graphs are given in Table I.

Table I. Slope of the ¹²⁹Xe Chemical Shift versus Protein Concentration Curves (see Fig. 5) for HisF in the Folded and Unfolded State as well as in the Folded State Bound to the Reaction Product ImGP

Sample	Slope (ppm/mM)
HisF folded	0.94 ± 0.03
HisF unfolded	0.55 ± 0.03
HisF ImGP	0.92 ± 0.03

Identification of xenon binding sites

On the basis of the existing spectral assignment, ¹H-¹⁵N HSQC experiments on the xenon-pressurized sample were used to identify and localize xenon-induced ¹H and/or ¹⁵N chemical shift changes by comparison with the xenon-free sample. Such chemical shifts sensitively indicate minor local structural changes of the protein induced by specific binding of xenon.14–18 In our experiments, the sample was equilibrated at 11 bar xenon pressure. The spectral assignment was based on the data published by Lipchock and Loria8 as well as our own work. The observed chemical shift changes were analyzed by calculating the so-called total xenon-induced chemical shift (Δ) from the xenon-induced ¹H and ¹⁵N chemical shifts as suggested by Gröger et al.15 The result of this analysis is shown in Figure 6. Note that the calculation of the combined chemical shifts, according to Schumann et al.,19 led to very similar results (data not shown). The horizontal line in Figure 6 indicates the average value of Δ plus its standard deviation. There are 28 amino acid residues with Δ-values beyond this line. These residues are considered to be strongly influenced by xenon binding. Ten of these residues are distributed more or less randomly at the protein surface. Six of them are hydrophobic, which probably explains their rather strong individual interaction with xenon. The other 18 amino acid residues are located in the neighborhood of three distinct cavities of the protein (see below). They were mapped on the crystal structure of HisF5 (Fig. 7). Furthermore, two amino acid residues (Ser201 and Asp183) exhibited severe line broadening in xenon-loaded samples, which led to their disappearance from the spectra.

Figure 6. Total xenon-induced chemical shift changes Δ observed in HisF as a function of xenon pressure. The average value plus one standard deviation is indicated as a horizontal line. Some of the residues with the highest Δ values are labeled.

Figure 7. Cavities (red) as calculated by HOLLOW in the crystal structure of HisF. Residues with large xenon induced chemical shift changes are shown in blue. The two red-marked residues Thr78 and Ser201 are putatively involved in ammonia transport and are strongly influenced by xenon as well. Top: Side view of the protein. The activity face comprises the C-terminal ends of the central barrel and the loops connecting the eight β-strands with the subsequent eight α-helices. The stability face comprises the remainder of the fold, including the loops connecting α-helices and subsequent β-strands. For a better view into the core of the barrel, one helix has been omitted. Bottom: Top view onto the active site. An interactive view is available in the electronic version of the article. Interactive View

We have determined internal cavities within the crystal structure of HisF using the computer program HOLLOW. Three cavities could be identified, which are surrounded by the majority of the strongly xenon-influenced amino acid residues (Fig. 7). Two cavities (cavity I and cavity II) are located on opposite sides of the PRFAR-binding groove at the activity face of HisF. Cavity III is located in the aforementioned central channel and is relatively close to the stability face of HisF. The largest chemical shift changes as well as the severe line broadening mentioned above were observed for amino acid residues surrounding cavity II (Table II).

Table II. Signals of Amino Acid Residues with Xenon-Induced Shifts Exceeding the Average Value Plus One Standard Deviation

Residues located at the walls of the three identified cavities are given in the upper three rows. Residues in parentheses are sequentially very close to amino acid residues located at the cavity walls. Asp183 and Ser201 are strongly broadened and, hence, “invisible” in the xenon-loaded sample, which prevents the measurement of the induced chemical shift. “Others” are the 10 xenon-sensitive amino acid residues distributed randomly over the sequence.
Cavity I	(Val56), Lys58, Arg59, Val79, Gly80
Cavity II	Leu170, Thr171, (Ile173), (Asp174), (Ser180), Asp183, Ala200, Ser201
Cavity III	Glu46, Thr78, Glu167, Gly166, Leu169
Others	Val12, Asp14, Val18, Val66, Glu67, Lys68, Ala204, Gly205, Val234, Leu241

It is known that xenon atoms are capable of binding even into hydrophobic cavities, which are somewhat smaller than a xenon atom. Such binding events require xenon-induced structural rearrangements of the amino acid residues surrounding the cavity resulting in cavity widening.15, 18 The overall volumes of the three cavities found to interact with xenon were determined using the web server CASTp (Table III). Their total volumes do all significantly exceed the volume of a single xenon atom. A xenon atom could be placed entirely into cavity II without inducing structural changes. In contrast, cavities I and III need to widen slightly to host a xenon atom due to their rather elongated shapes. Note that water molecules are present in the cavities (Table III) according to the crystal structure.5 The walls of the three cavities are formed predominantly by hydrophobic residues but contain also a few hydrophilic amino acids. Xenon is, however, able to bind into amphiphilic regions despite its clear preference for hydrophobic environments.20

Table III. Dimensions of the Detected Cavities and Their Respective Water Occupancya

	Surface (Å^2)	Volume (Å^3)	Water occupancy
a The water accessible surface and the volume of the cavity was determined as described in Materials and Methods.
Cavity I	154	119	2
Cavity II	102	83	3
Cavity III	114	76	1
Xenon	61	45	-

Discussion

We have shown in this study that the protein HisF from the hyperthermophilic bacterium T. maritima specifically binds the hydrophobic noble gas xenon. Significantly xenon-influenced amino acid residues are located preferentially in the neighborhood of three distinct cavities, which were identified by a computational analysis of the crystal structure. It should be noted that the effects measured for HisF were smaller than those observed for proteins with engineered xenon-binding sites,14 such as the I14A mutant of HPr from S. carnosus.15, 18 This can be explained by the different sizes of the cavities. Point mutations, such as in HPr(I14A) result in smaller cavities than the ones found in wild-type HisF whose volumes are large enough to readily accommodate a xenon atom (cf. Table III). Xenon binding to HisF should consequently result in less pronounced ¹H and ¹⁵N chemical shifts, which are caused by structural rearrangements.18 Intramolecular cavities are necessary to facilitate molecular motions and conformational changes of proteins.21 In the case of HisF, conformational changes could be necessary for substrate binding, catalysis, and/or signaling between HisF and the glutaminase HisH.

Molecular dynamics simulations indicate that the highly conserved residues Thr78, Ser101, and Ser201 serve as “ammonia-relays” during its transport through the channel.22, 23 The side chain of Thr78 points into cavity III, Ser101 is located at the bottom of the ligand binding cleft, and Ser201 is part of the wall of cavity II. The detected interaction of NH₃ with Thr78 and the four-fold decrease of the catalytic turnover in the Thr78Met mutant as detected in our study are consistent with the idea that Thr78 is involved in ammonia transport through the channel. Only one among the 6 amino acids, which are sensitive to the exchange of NH₃/NH by Na⁺ is charged (Glu71). This indicates that the experiment indeed detects interactions of uncharged NH₃ with HisF.

The signal of Ser201 in the neighborhood of cavity II disappears in the xenon-loaded sample, whereas data for Ser101 are missing due to signal overlap. By allowing molecular motion, the xenon binding sites might be involved in the transport of NH₃ through the centre of the barrel. A similar proposal was made for the transport of ligands in myoglobin involving several xenon-binding sites.24

The inhibition of histidine biosynthesis is a potential therapeutic concept because only bacteria, fungi, and plants possess this anabolic pathway. Some applications of this principle have already been developed for specific cases.25, 26 So far, however, there are no drugs available targeting the enzymes of histidine biosynthesis. The natural substrate of HisF binds to the active site via two phosphate moieties. For drug development, these moieties would at least have to be replaced by other charged residues because phosphate-containing molecules are rather poor candidates for drug design. However, an alternative way to inhibit the enzyme's function may be hindering a possibly essential conformational change during catalysis. Blocking the newly discovered cavities I and II by small molecules as has been accomplished already in the case of MBP14 may fulfill this purpose.

Materials and Methods

¹H-¹⁵N correlation spectroscopy

Experiments were recorded at 298 K. Xenon-loaded samples were measured on a Bruker DMX 500 and NH₄Cl samples on a BRUKER AVANCE 800 with a TCI cryogenic probe. Pulse programs for the ¹H-¹⁵N correlation spectra were hsqcetf3gpsi and trosyf3gpphsi19 for xenon-loaded and NH₄Cl-containing samples, respectively. For the proton and nitrogen dimensions of xenon-loaded samples, spectra were recorded with 8012 Hz frequency width/2048 data points and 2000 Hz frequency width/512 data points, respectively. NH₄Cl samples were acquired with 11,161 Hz frequency width/2096 data points in the proton dimension and 2920 Hz frequency width/data 512 points in the nitrogen dimension. The “total xenon induced shift” Δ at 11 bar xenon gas pressure was calculated according to Gröger et al.15 Spectra assignment was based on the data published by Lipchock and Loria8 as well as our own work (to be published).

Samples contained 1.1 (xenon-loaded) and 1.2 mM (NH₄Cl) ¹⁵N-labeled HisF in 50 mM Tris/HCl, pH 7.5 containing 10% D₂O and 60 μM DSS. Stock solutions (5M) of NH₄Cl and NaCl (both, p.a. Merck KGaK, Darmstadt, Germany) pH 7.5 were used to prepare samples with 0, 10, 50, 200, 400, and 578 mM NH₄Cl, and one sample with 200 mM NaCl and 200 mM NH₄Cl. Differences between the ¹⁵N chemical shift of the 400 mM NH₄Cl sample and the sample containing 200 mM NaCl and 200 mM NH₄Cl were calculated. Two times the experimental error was the cut off for significant ammonia-induced chemical shift changes.

Data processing

All spectra were processed with the topspin package (Bruker BioSpin) and further analyzed with AUREMOL.27 Proton chemical shifts were referenced to DSS, whereas ¹⁵N chemical shifts were referenced indirectly using the procedure described by Wishart et al.28 and Markley et al.29

¹²⁹Xe chemical shift measurements

Spectra were recorded with a broadband probe using direct excitation with a relaxation delay of 300 s and at least 52 scans. Samples for HisF were prepared in 50 mM Tris/HCl pH 7.5 with 10% D₂O and 60 μM DSS and contained additionally 6M guanidinum chloride (Gerbu Biotechnik, GmbH, Germany) or 13 mM ImGP (purity 90%, Toronto Research Chemicals, Canada) for unfolded and ligand-bound HisF, respectively. Concentrations were determined before and after the NMR measurements and the mean value was calculated (1130, 810, 562, 330, and 220 μM HisF-folded; 479, 240, and 117 μM HisF-unfolded; 890, 445, and 220 μM HisF-ImGP). ¹²⁹Xe chemical shifts in protein solutions were referenced relative to the ¹²⁹Xe chemical shift in the respective buffer.

High-pressure equipment

Samples were pressurized using a home-made high pressure apparatus30, 31 equipped with a sapphire tube (Saphikon, USA) similar to the one used by Baumer et al.32 Samples were filled with natural abundance xenon gas (Xenon 4.0 > 99.99%, Linde Gas AG, Germany). After all ¹²⁹Xe NMR measurements, the samples were tested for tightness (i.e., constant gas content) to avoid erroneous chemical shift changes because of variations of the xenon pressure.

UV measurements

The concentration of HisF was measured by UV spectroscopy using a molar extinction coefficient at 280 nm of 11460 M⁻¹ cm⁻¹ for native and 10810 M⁻¹ cm⁻¹ for the unfolded protein, respectively, as determined from the amino acid sequence.33, 34

Cavity calculation

HOLLOW35 was used to calculate the inner and outer surface of HisF. Cavity volumes and solvent accessible surface were calculated using the web server CASTp36 with a rolling sphere of 1.4 Å. PyMOL37 was used for the visualization of data.

Generation and cloning of hisF-T78M

The mutation T78M was introduced into hisF by the megaprimer method using pET11c-thisF4 as template. The resulting hisF-T78M gene was cloned into pET28a using NdeI and BamHI sites.

Heterologous gene expression and protein purification

For ¹⁵N labeling of HisF, E. coli BL21(DE3) (Stratagene) cells transformed with pET11c-hisF were grown in 1 L of minimal medium containing 150 μg/mL ampicillin. Minimal medium contained 7.5 g Na₂HPO₄·H₂O, 3 g KH₂PO₄, 0.5 g NaCl, 0.25 g MgSO₄·7H₂O, and 0.014 g CaCl₂·2H₂O. It was autoclaved and then supplemented with 1 g ¹⁵NH₄Cl, 10 g glucose, and 10 mL trace elements stock solution containing 50 mg EDTA, 20 mg Fe(II)SO₄·7H₂O, 10 mg ZnSO₄·7H₂O, 3 mg MnCl₂, 30 mg H₃BO₃, 20 mg CoCl₂·6H₂O, 1 mg CuCl₂·2H₂O, 2 mg NiCl₂·6H₂O, and 3 mg Na₂MoO₄·2H₂O. The recombinant protein product was purified from the soluble fraction of the crude extract by heat precipitation of the host proteins (20 min at 70°C) followed by ion exchange chromatography. For this purpose, the extract was loaded onto a MonoQ column (HR 16/10, 20 mL, Pharmacia), which had been equilibrated with 50 mM Tris/HCl pH 7.5. The column was washed with the equilibration buffer, and bound protein was eluted by applying a linear gradient of 0–1.5M NaCl. Fractions with pure protein were pooled and dialyzed extensively against 50 mM Tris/HCl pH 7.5. Unlabeled HisF was expressed in Bl21(DE3) cells in LB medium, and purified as described above.

For expression of nonlabeled HisF-T78M, E. coli BL21(DE3)Rosetta2 cells transformed with pET28a-hisF-T78M were grown in 1 L of LB-medium containing 75 μg/mL kanamycin. Expression from pET28a resulted in the addition of an N-terminal hexa-histidine tag to the recombinant protein. The supernatant containing His₆-tagged protein was loaded onto a nickel sepharose column (HisTrap FF crude 5 mL, GE Healthcare), which had been equilibrated with 50 mM potassium phosphate, 300 mM NaCl, 10 mM imidazole, pH 7.5. The column was washed with the equilibration buffer, and the bound His₆-tagged protein was eluted by applying a linear gradient of 1–500 mM imidazole. Fractions with pure protein were pooled and dialyzed extensively against 50 mM potassium phosphate pH 7.5.

HisH was produced in E. coli strain W3110ΔtrpEA2 containing the helper plasmid pDMI,1 as described for HisA from T. maritima.38 The recombinant protein product was purified from the soluble fraction of the crude extract by heat precipitation of the host proteins (20 min at 70°C), followed by ion exchange chromatography. To this end, the extract was loaded onto a MonoQ column (HR 16/10, 20 mL, Pharmacia), which had been equilibrated with 50 mM potassium phosphate (pH 7.5). The column was washed with the equilibration buffer, and bound protein was eluted by applying a linear gradient of 0–1.5M NaCl. Fractions with pure protein were pooled and dialyzed extensively against 50 mM potassium phosphate (pH 7.5).

Steady state kinetics

The glutamine- and ammonia-dependent cyclase activities of the ImGP synthase bienzyme (HisH:HisF complex) were measured at 25°C in Tris/Acetat pH 8.0 as previously described,4 using PRFAR produced in situ from ProFAR by HisA from T. maritima as the substrate. To assure that all HisF subunits are engaged in a complex, the HisH and HisF proteins were mixed in a 2:1 molar ratio before the measurements. The values for K and V_max were deduced from initial velocity measurements performed in the presence of various concentrations of PRFAR. The turnover number k_cat was calculated by dividing V_max by the total concentration of HisF [E₀]. Given values are the averages of two independent measurements, which yielded similar numbers.

Acknowledgements

The authors thank Michaela Haeger for cloning of pET11c-hisF-T78M. Thanks are further due to Ms. Annett Bachmann and Ms. Renate Schulze (Dresden) for carefully proofreading the manuscript.