X-ray structure and characterization of carbamate kinase from the human parasite Giardia lamblia

The three-dimensional structure of carbamate kinase from the human parasite G. lamblia has been determined at 3 Å resolution. Steady-state kinetic parameters have been determined and dsRNA-silencing experiments indicate that the enzyme is essential for the survival of G. lamblia.

Abstract

Carbamate kinase catalyzes the reversible conversion of carbamoyl phosphate and ADP to ATP and ammonium carbamate, which is hydrolyzed to ammonia and carbonate. The three-dimensional structure of carbamate kinase from the human parasite Giardia lamblia (glCK) has been determined at 3 Å resolution. The crystals belonged to the monoclinic space group P2₁, with unit-cell parameters a = 69.77, b = 85.41, c = 102.1 Å, β = 106.8°. The structure was refined to a final R factor of 0.227. The essentiality of glCK together with its absence in humans makes the enzyme an attractive candidate for anti-Giardia drug development. Steady-state kinetic rate constants have been determined. The k _cat for ATP formation is 319 ± 9 s⁻¹. The K _m values for carbamoyl phosphate and ADP are 85 ± 6 and 70 ± 5 µM, respectively. The structure suggests that three invariant lysine residues (Lys131, Lys216 and Lys278) may be involved in the binding of substrates and phosphoryl transfer. The structure of glCK reveals that a glycerol molecule binds in the likely carbamoyl phosphate-binding site.

1. Introduction

Carbamate kinase (CK; EC 2.7.2.2; ATP:carbamate phosphotrans­ferase) from the arginine dihydrolase pathway catalyzes the reversible conversion of carbamoyl phosphate and ADP to ATP and ammonium carbamate, which is hydrolyzed to ammonia and carbon­ate. The three-step arginine dihydrolase pathway, involving arginine deiminase, ornithine transcarbamoylase and CK, is used by a number of microorganisms, including Giardia lamblia, to generate ATP (Biagini et al., 2003 ▶; Knodler et al., 1998 ▶). G. lamblia is the causative agent of the human intestinal disease giardiasis (for a review, see Adam, 2001 ▶), the most common cause of waterborne outbreaks of diarrhea in the United States. Infections are prevalent in impoverished countries, leading to much suffering and mortality. Because the arginine dihydrolase pathway is absent in humans and other high eukaryotes, the enzymes of this pathway are potential targets for drug development. Arginine deiminase and ornithine transcarbamoylase have been shown to be among the major proteins released into the medium after brief interaction of G. lamblia with human intestinal epithelial cells, underscoring the importance of the arginine catabolism pathway in host colonization by the parasite (Palm et al., 2003 ▶). The G. lamblia CK (glCK) was found to be the most active of the arginine dihydrolase enzymes, with a maximal activity in cell-free extracts that was fivefold to tenfold greater than those of arginine deiminase and ornithine transcarbamoylase (Schofield et al., 1992 ▶). Northern hybridization experiments revealed that the glCK gene is primarily expressed in G. lamblia trophozoites and that expression is significantly decreased during encystation (Minotto et al., 1999 ▶). The glCK gene is also one of eight genes whose expressions are up-regulated in neomycin-selected G. lamblia cell lines (Su et al., 2007 ▶), suggesting its importance under stress conditions. The G. lamblia arginine deiminase and ornithine transcarbamoylase have recently been purified and functionally characterized in our laboratory (Galkin et al., 2009 ▶; Li et al., 2009 ▶). Here, we show that glCK is essential for the viability of G. lamblia WB trophozoites and we also report the purification and structural and functional characterization of the enzyme. The essentiality of glCK together with its absence in humans makes the enzyme an attractive candidate for anti-Giardia drug development.

2. Experimental procedures

2.1. Gene cloning, protein expression and purification

Trophozoites of G. lamblia isolate WB were grown as described previously (Wieder et al., 1983 ▶). The genomic DNA was isolated using the DNA STAT kit (Stratagene). The gene encoding glCK was amplified using Pfu Turbo DNA polymerase (Stratagene), genomic DNA and 5′-end and 3′-end primers. The PCR product was inserted into the pDEST-HisMBP expression vector as described previously (Nallamsetty et al., 2005 ▶). For protein production, glCK was expressed in Escherichia coli strain BL21 (DE3) Star as a maltose-binding protein (MBP) fusion product. Cells were grown in Overnight Express Instant TB autoinduction medium (Novagen) for 20 h at 303 K. The cells were lysed by sonication and the soluble fraction was chromatographed on an Ni–NTA affinity column. After elution and concentration, the glCK MBP-fusion protein was chromatographed on a preparative Sephacryl 100 size-exclusion column. MBP was removed by treatment of the fusion protein with tobacco etch virus (TEV) protease (Nallamsetty et al., 2004 ▶) at a molar ratio of 60:1, followed by chromatography on an Ni–NTA affinity column to remove MBP and uncleaved protein. The protein was further purified by ion-exchange chromatography on Source 15Q columns (Amersham Biosciences). Fractions containing protein were collected, dialyzed against 50 mM Tris–HCl pH 7.5 and 50 mM NaCl and con­centrated to 45 mg ml⁻¹. Protein integrity and purity was assessed by polyacrylamide gel electrophoresis in the presence of SDS. The oligomeric state was measured by analytical size-exclusion chromatography on an ÄKTA Purifier 10 using a Superdex-200 HR 10/30 column (Amersham Biosciences).

2.2. Functional knockdown of the CK gene in G. lamblia

The glCK gene was inserted into the plasmid between two opposite tetracycline-inducible Ras-related nuclear protein (ran) promoters as described previously (Touz et al., 2004 ▶). Accordingly, the dsRNA was synthesized from a DNA template upon induction with tetracycline at a final concentration of 10 µg ml⁻¹. Stable transfections of G. lamblia trophozoites were produced by electroporation of circular plasmids containing a puromycin-resistance gene. Approximately 10⁷ trophozoites were resuspended in 0.3 ml medium, mixed with 10 µg DNA and incubated on ice for 5 min. Cells were electroporated in a 0.4 cm cuvette with an ECM 600 (BTX, San Diego, California, USA) set to 350 V, 1000 µF and 720 Ω and transferred to 15 ml medium in a glass tube after 10 min on ice. After overnight incubation without puromycin, cultures were chilled on ice and additional media and drug were added to a final volume of 20 ml and 100 µM puromycin. Cells were then distributed into a 96-well plate and sealed in an anaerobic environment.

2.3. Steady-state kinetics

For the conversion of ATP and ammonium carbamate to ADP and carbamoyl phosphate, reaction solutions (1 ml) initially contained ATP at varied concentration (0.5-fold to fivefold K _m) and 30 mM ammonium carbamate (or ammonium carbamate at varied concentrations and 3.5 mM ATP), glCK, 3 mM PEP, 11.5 mM MgCl₂, 0.2 mM NADH, 10 U lactate dehydrogenase and 10 U pyruvate kinase in 50 mM Tris–HCl pH 7.5 at 298 K. The progress of the reaction was monitored at 340 nm (Δ∊ = 6.2 mM ⁻¹ cm⁻¹). The initial velocities (V ₀) measured as a function of the concentration of the varied substrate, [S], were fitted with KinetAsyst to the equation V ₀ = V _max[S]/(K _m + [S]) to obtain the maximum velocity (V _max) and the Michaelis constant (K _m). k _cat was calculated from the ratio of the V _max and the enzyme concentration.

For the conversion of ADP and carbamoyl phosphate to ATP and ammonium carbamate, reaction solutions (1 ml) initially contained ADP at varied concentrations (0.5-fold to fivefold K _m), 2 mM carbamoyl phosphate (or carbamoyl phosphate at varied concentration and 5 mM ADP), glCK, 2 mM d-glucose, 200 µM NADP, 10 U hexokinase, 5 U glucose-6-phosphate dehydrogenase, 0.002%(w/v) bovine serum albumin and 30 mM MgCl₂ in 20 mM Tris–HCl pH 8.3 at 310 K. The reaction progress was monitored at 340 nm (Δ∊ = 6.2 mM ⁻¹ cm⁻¹). The k _cat and K _m values were calculated from the initial velocity data as described above and are given in Table 1 ▶.

Table 1. Steady-state kinetic parameters.

Substrate	Km (µM)	kcat (s−1)	kcat/Km (s−1  M−1)
ATP†	35 ± 2	—	—
Carbamate‡	43 × 103 ± 11 × 103	43 ± 7	1.0 × 103
Carbamoyl phosphate§	85 ± 6	319 ± 9	3.8 × 106
ADP¶	70 ± 5	319 ± 9	4.6 × 106

^†Reactions contained 30 mM ammonium carbamate and 11.5 mM MgCl₂. Because the carbamate was not saturating (i.e. tenfold K _m) in the reaction mixture, the value of the observed k _cat of 16.1 ± 0.3 s⁻¹ is less than the actual k _cat and is therefore not reported in the table. The use of high concentrations of ammonium carbamate caused inhibition.

^‡Reactions contained 3.5 mM ATP, 15 mM MgCl₂ and a variable amount of NH₄Cl and KHCO₃. NH₄Cl and KHCO₃ were added as a mixture that had been kept at ambient temperature for 20 min to allow equilibration with carbamate. The carbamate concentration was calculated using the published equilibrium constant (Marshall & Cohen, 1966 ▶).

^§Reactions contained 5 mM ADP and a variable amount of carbamoyl phosphate.

^¶Reactions contained 2 mM carbamoyl phosphate and a variable amount of ADP.

The competitive inhibition constants for adenosine, adenosine 5′-­monosulfate and AMPPNP were determined by fitting the initial velocities measured for the ATP-producing reaction in the presence of varied concentrations of the compound examined. The initial velocity data were analyzed using the equation V ₀ = V _max[S]/{K _m(1 + [I]/K _i) + [S]}, where K _i is the inhibition constant and [I] is the concentration of the compound.

2.4. Crystallization, data collection and structure determination

The following commercially available crystallization kits were screened: Crystal Screen, Crystal Screen 2, PEG/Ion and PEG/Ion 2 from Hampton Research, Wizard I and Wizard II from Emerald BioSystems and JCSG Core Suite I from Qiagen. The PEG/Ion screen yielded the best crystals. Optimized crystals of glCK were obtained at 293 K by the vapor-diffusion method in hanging drops. The protein solution was mixed with an equal volume of mother liquor containing 20% PEG 3350 and 0.2 M ammonium citrate and equilibrated against the mother-liquor reservoir. The crystals were transferred to mother liquor containing 20% glycerol and flash-cooled at 160 K. Diffraction data were acquired using an R-AXIS IV⁺⁺ image-plate detector mounted on a Rigaku rotating-anode MicroMax-007 X-ray generator (Rigaku MSC Inc.). The crystals diffracted X-rays to a resolution of 3.0 Å. Data processing was carried out using CrystalClear v.1.3.6 (Rigaku MSC Inc.). The statistics of data collection are provided in Table 2 ▶.

Table 2. X-ray data-collection and refinement statistics.

Values in parentheses are for the highest resolution shell.

Space group	P21
Crystal solvent content (%)	43
No. of molecules in asymmetric unit	4
Data collection
Unit-cell parameters (Å, °)	a = 69.77, b = 85.41, c = 102.1, β = 106.8
Resolution range (Å)	20–3.0 (3.1–3.0)
No. of observations	103189
No. of unique reflections	20419
Completeness (%)	88.4 (87.0)
Rmerge†	0.095 (0.193)
Refinement statistics
No. of refined reflections	18918 (1193)
No. of reflections in Rfree set	998 (70)
Resolution range (Å)	10–3.0 (3.07–3.0)
No. of residues	1161
Rcryst‡	0.227 (0.239)
Rfree§	0.283 (0.381)
R.m.s. deviations
Bonds (Å)	0.018
Angles (°)	2.1
Backbone dihedral angle distribution (%)
Allowed	95.3
Generously allowed	4.7
Disallowed	0.0
〈B〉 (Å2)	38.4

^† R _merge = .

^‡ R _cryst = − , where F _obs and F _calc are the observed and calculated structure factors, respectively

^§ R _free is computed for 5% of reflections that were randomly selected and omitted from the refinement.

The crystal structure of glCK was determined by molecular-replacement techniques using the program Phaser (McCoy et al., 2005 ▶), with the Pyrococcus furiosus CK (PfCK) structure (Ramon-Maiques et al., 2000 ▶; PDB code 1e19) as a search model. The difference Fourier maps indicated some alternative polypeptide tracing, which was rebuilt on a graphics workstation using the program O (Kleywegt & Jones, 1999 ▶). Structure refinement was carried out using CNS (Brünger et al., 1998 ▶) and REFMAC (Mur­shudov et al., 1997 ▶). The resulting models were inspected and modified on a graphics workstation. Given the low resolution of the diffraction, group temperature factors were used (with two groups per amino-acid residue) and no water molecules were included in the model. Refinement statistics are provided in Table 2 ▶.

3. Results and discussion

3.1. Expression, purification and kinetic properties of glCK

The N-terminally His-tagged MBP-fusion glCK was expressed in soluble form in E. coli and remained soluble after digestion with TEV protease and removal of the MBP. Analytical size-exclusion chromatography of purified glCK indicated that the protein primarily exists as a homodimer in solution. ATP/AMP analogs were found to be weak binding inhibitors, with K _i values of 8.5 mM for adenosine, 5 mM for adenosine 5′-monosulfate and 0.8 mM for AMPPNP. Although AMPPNP is a rather weak inhibitor of glCK, it is the best inhibitor reported to date.

3.1.1. glCK essentiality test

Previous experiments have confirmed that the CK gene is functional in G. lamblia trophozoites and that glCK is expressed in vivo (Minotto et al., 1999 ▶; Schofield et al., 1992 ▶). To further validate glCK as a potential drug-target candidate, we tested the essentiality of the glCK gene by using the RNAi gene-silencing technique (Bernstein et al., 2001 ▶), which has recently been adopted for G. lamblia trophozoites (Touz et al., 2004 ▶). We have previously used this approach to validate the G. lamblia arginine deiminase and fructose-1,6-bisphosphate aldolase as potential drug targets (Li et al., 2009 ▶; Galkin et al., 2007 ▶). The G. lamblia trophozoites were transfected with the glCK gene as described above. No viable trophozoites were observed after 10 d of cultivation. In a control experiment using a nonessential gene encoding the adaptor protein involved in protein trafficking (Touz et al., 2004 ▶), viable trophozoites were observed after transfection. The glCK dsRNA triggered transcriptional inhibition and glCK gene silencing, which was lethal for all trophozoites, consistent with the essential role that the enzyme plays in the survival of the parasite.

3.2. Overall structure of glCK

The glCK crystal contains four protein molecules in the asymmetric unit (labeled A, B, C and D in the PDB entry) that pack into two homodimers consistent with the oligomeric form in solution. The dimer is depicted in Fig. 1 ▶. Two archeal CKs with known structures, PfCK and Aeropyrum pernix CK (apCK; K. Shimizu & N. Kunishima, unpublished work; PDB code 2e9y), and a bacterial enzyme from Entero­coccus faecalis (EfCK) also form dimers (Durbecq et al., 1997 ▶; Marina et al., 1999 ▶; Ramon-Maiques et al., 2000 ▶).

Figure 1.

Overall fold of glCK. A ribbon-diagram representation of the protein dimer is shown. Each subunit is represented in a different color. The glycerol molecules in the active sites are shown in red.

The model of the four glCK molecules in the asymmetric unit includes 1161 of the 1264 amino-acid residues. A glycerol molecule originating from the cryoprotecting solution is located in each active site in a position where the carbamoyl phosphate is expected to bind (Fig. 2 ▶ a). Missing from each of the four molecules are residues that have no associated electron density: the N-terminal residue remaining after TEV protease cleavage (Gly) and the authentic N-­terminal Met residue. In addition, no electron density is associated with the following surface residues: 18–19, 135–160 and 272–273 in molecule A, 138–152, 244–245 and 271–272 in molecule B, 138–161 and 247–250 in molecule C, and 140–152, 243–249 and 271–272 in molecule D. These residues were omitted from the final model.

Figure 2.

View of the glCK active site. (a) The σ_A-weighted 2F _o − F _c difference Fourier electron-density map in the vicinity of the glycerol molecule (labeled GL) calculated omitting the ligand from the model. Atomic colors are as follows: carbon, yellow; oxygen, red; nitrogen, blue; sulfur, green. (b) The lysine and aspartic acid cluster in the vicinity of the glycerol molecule. Atomic colors are as follows: carbon, gray (glCK) and green (glycerol, labeled GL); oxygen, red; nitrogen, blue.

The glCK molecule folds into a three-layer α/β/α sandwich with parallel open β-sheets surrounded by α-helices, forming a modified ‘Rossmann fold’ as observed in the structures of the related bacterial CK (Marina et al., 1999 ▶). The glCK subunit consists of N-terminal (residues 2–228) and C-terminal (residues 229–316) domains. As in other CKs, the N-terminal domains are positioned in the center of the dimer, thus mediating the intersubunit interface. A DALI (Holm & Sander, 1993 ▶) search identified three CK structures similar to glCK: pfCK (PDB code 1e19; Ramon-Maiques et al., 2000 ▶; Z score of 39.4; r.m.s.d. of 1.7 Å for 282 aligned C^α atoms that share 50% sequence identity), efCK (PDB code 1b7b; Marina et al., 1999 ▶; Z score of 38.2; r.m.s.d. of 1.5 Å for 278 aligned C^α atoms that share 43% sequence identity) and apCK (PDB code 2e9y; K. Shimizu & N. Kunishima, unpublished work; Z score of 39.2; r.m.s.d. of 1.7 Å for 282 aligned C^α atoms that share 40% sequence identity).

3.3. Active site of glCK

The glCK active site is of special interest for the development of inhibitors which might lead to therapeutic agents. The large crevice between the N- and C-terminal domains of pfCK hosts the site for Mg²⁺-ADP (Ramon-Maiques et al., 2000 ▶). A sulfate ion binds at the bottom of the active-site crevice of the efCK structure, occupying a site that is expected to bind the γ-phosphate of ATP or the phosphate group of carbamoyl phosphate (Marina et al., 1999 ▶). A strong peak was found in the glCK electron-density map in close proximity to the postulated efCK sulfate ion. Because of the shape of the electron density and the lack of sulfate in the crystallization solution of glCK, the electron density was modeled as a glycerol molecule (Fig. 2 ▶). The environment of both the efCK sulfate ion and the glCK glycerol includes the backbone amides of two conserved glycine residues (Gly13 and Gly56 in the glCK sequence) as well as the side chains of five more conserved residues (Lys131, Lys216, Lys278, Asp215 and Asp217; Fig. 2 ▶ b). The bound sulfate in efCK was proposed to occupy the site of the ATP γ-phosphate, the group that is transferred to generate carbamoyl phosphate (Marina et al., 1999 ▶). The glCK structure reveals that this site also accommodates a glycerol molecule, a noncharged polar moiety with the potential to donate or accept hydrogen bonds. Superposition of the glCK structure and that of the pfCK–Mg²⁺-ADP complex suggests that if a carbamoyl phosphate binds in the glycerol site its phosphate may be positioned in line to be transferred to the ADP β-phosphate and to generate a pentavalent phosphorus intermediate/transition state (Fig. 3 ▶). This transient adduct would have a trigonal bipyramidal geometry with the phosphoryl donor and acceptor groups located at apical positions (Knowles, 1980 ▶). The model does not require adjustment of the protein structure and suggests that three invariant lysine residues (Lys131, Lys216 and Lys278) may be involved in substrate binding and phosphoryl transfer (Fig. 3 ▶). Lys278 may interact with the ADP β-phosphate to position the substrate for optimal phosphate transfer. Lys216 may interact with the carbamoyl phosphate to counter a negatively charged pentavalent phosphorus intermediate, which has also been suggested for the equivalent pfCK Lys215 (Ramon-Maiques et al., 2000 ▶). It is noteworthy that the correct positioning of Lys216 for catalysis is ensured by two invariant aspartate residues (Asp215 and Asp217). When the equivalent aspartate residues of efCK (Asp208 and Asp210) were replaced by alanine residues, the activity of the mutants was reduced by 1000-fold compared with the wild-type enzyme (Marina et al., 1999 ▶). The electrostatic network that supports the reaction center is further extended: the carboxyl group of Asp217 interacts with the amino groups of both the conserved Lys278 and Lys216 and the carboxyl group of Asp215 forms hydrogen bonds to the backbone amides of Asp217 and Leu218. The third invariant active-site lysine residue, Lys131, may interact with the carbamoyl O atom of carbamoyl phosphate and fine-tune the orientation of the substrate for efficient phosphate transfer.

Figure 3.

Stereoscopic view of the glCK active site with modeled Mg²⁺-ADP and carbamoyl phosphate. The Mg²⁺-ADP was modeled by analogy to that observed in the pfCK structure (Ramon-Maiques et al., 2000 ▶). The same color scheme is used for carbon, oxygen and nitrogen as in Fig. 2 ▶(b); P atoms are colored orange.

Comparison of glCK with pfCK revealed intriguing conformational changes associated with the binding of the adenine moiety of ADP (Fig. 4 ▶). ADP binding promotes loop closure and a shift in the position of a tyrosine side chain so that the aromatic ring stacks against the adenine ring. The second difference is the disorder of the loop residues 272–273 in the absence of ADP.

Figure 4.

Superposition of glCK (green) and pfCK (gray) showing the environment of the adenine group of ADP. A tyrosine-carrying loop adopts a more closed conformation in pfCK than in glCK such that the tyrosine residue stacks against the adenine ring. A loop that flanks the other side of the adenine ring is partially disordered in the crystal structure of glCK, which does not contain ADP.

Site-directed mutagenesis experiments together with structures of liganded glCK will provide better insight into the CK reaction mechanism and strategies for inhibitor development.

Supplementary Material

PDB reference: carbamate kinase, 3kzf