Crystal Structure of Butyrate Kinase 2 from Thermotoga maritima, a Member of the ASKHA Superfamily of Phosphotransferases

Abstract

The enzymatic transfer of phosphoryl groups is central to the control of many cellular processes. One of the phosphoryl transfer mechanisms, that of acetate kinase, is not completely understood. Besides better understanding of the mechanism of acetate kinase, knowledge of the structure of butyrate kinase 2 (Buk2) will aid in the interpretation of active-site structure and provide information on the structural basis of substrate specificity. The gene buk2 from Thermotoga maritima encodes a member of the ASKHA (acetate and sugar kinases/heat shock cognate/actin) superfamily of phosphotransferases. The encoded protein Buk2 catalyzes the phosphorylation of butyrate and isobutyrate. We have determined the 2.5-Å crystal structure of Buk2 complexed with (β,γ-methylene) adenosine 5′-triphosphate. Buk2 folds like an open-shelled clam, with each of the two domains representing one of the two shells. In the open active-site cleft between the N- and C-terminal domains, the active-site residues consist of two histidines, two arginines, and a cluster of hydrophobic residues. The ATP binding region of Buk2 in the C-terminal domain consists of abundant glycines for nucleotide binding, and the ATP binding motif is similar to those of other members of the ASKHA superfamily. The enzyme exists as an octamer, in which four disulfide bonds form between intermolecular cysteines. Sequence alignment and structure superposition identify the simplicity of the monomeric Buk2 structure, a probable substrate binding site, the key residues in catalyzing phosphoryl transfer, and the substrate specificity differences among Buk2, acetate, and propionate kinases. The possible enzyme mechanisms are discussed.

Butyrate kinases 1 and 2 are the two probable butyrate kinases from the thermophilic and fermentative organism Thermotoga maritima, which grows between 50°C and 90°C in geothermal heated marine sediment and metabolizes carbohydrates (22, 32). Butyrate kinase 2 (Buk2) can catalyze the phosphorylation of butyrate and isobutyrate, valerate, 2-methylbutyrate, and, to a lesser extent, propionate and isovalerate (8). By analogy with other butyrate kinases in the butyrate fermentation pathway (18), Buk2 (EC 2.7.2.7; 375 amino acids, 42 kDa), encoded by TM1756 or buk2 from T. maritima, may catalyze the phosphorylation of butyrate/isobutyrate to butyryl phosphate/isobutyryl phosphate, which is then converted to butyryl coenzyme A (butyryl-CoA)/isobutyryl-CoA by phosphotransbutyrylase, as follows: ATP + isobutyrate ⇆ ADP + isobutyryl phosphate; isobutyryl phosphate + CoA ⇆ isobutyryl-CoA + phosphate.

Butyrate kinases exist in many fermentative species of the bacterial and archaeal families. In the well-characterized species Clostridium acetobutylicum, a predominant biochemical pathway is the fermentation of butyric acid and butanol/acetone, in which butyryl-CoA is converted to butyric acid through the reactions running in the reverse direction (11, 18). The term fermentation was first defined as life in the absence of oxygen by Pasteur, who discovered the fermentation of sugars to butyric acid in 1861. Fermentations are classified according to main fermentation end products, and butyrate fermentation is one of several major fermentations (18). A sugar such as glucose is the common starting substrate of fermentation. Unlike yeast, which can digest only sugar, C. acetobutylicum and many other clostridia can digest whey, sugar, starch, lignin, cellulose fiber, and other biomass. During the early growth phase or the acid-producing phase of the fermentation by C. acetobutylicum, butyrate and acetate are the main metabolic products, which result in a decrease in the pH of the culture medium. As the culture continues to grow, a significant shift in the solvent-producing phase from butyrate to n-butanol/acetone production can be observed, and n-butanol/acetone become the main products of the fermentation (18). The species C. acetobutylicum has been used on an industrial scale for the synthesis of n-butanol and acetone from molasses, and n-butanol/acetone fermentation used to rank second only to ethanol fermentation by yeast in its scale of production and was one of the largest biotechnological processes known (14). The butyrate kinase from C. acetobutylicum can catalyze the phosphorylation of several different substrates (butyrate, valerate, isobutyrate, propionate, isovalerate, and vinyl acetate) (20), and several butyrate kinase genes from C. acetobutylicum variants have been identified through cloning studies (21, 39). The butyrate kinases from Bacillus subtilis and Enterococcus faecalis are not characterized, and insights on the butyrate biochemical pathways can be gained by examining the operons containing butyrate kinase genes. The butyrate biochemical pathways from B. subtilis (12, 25), E. faecalis (40), and T. maritima (32) may be involved in the catabolism of branched-chain amino acids or the reversible conversion of branched-chain amino acids into the corresponding fatty acids. Besides the above-mentioned butyrate-producing species, a diverse range of species of bacteria ferment butyrate in the human colon, and the fermentation product butyrate has been implicated in prevention of colon cancer (2, 6, 29).

Sequence studies suggest that the encoded protein, Buk2, belongs to the ASKHA superfamily of phosphotransferases, all members of which share a common characteristic five-stranded β sheet occurring in both the N- and C-terminal domains (5, 19). Buk2 (13) is closely related to acetate kinase and propionate kinase, whose structures have been determined recently (5, 35). The sequence identity between acetate kinase from Methanosarcina thermophila and Buk2 is only 24% of 375 residues of Buk2 (90/375), and the sequence identity between propionate kinase from Salmonella enterica serovar Typhimurium and Buk2 is only 19% (72/375). While the general folds and the reaction mechanisms of Buk2, acetate kinase, and propionate kinase are thought to be similar, the reaction mechanism of acetate kinase has been controversial for decades. On the one hand, it is suggested that the phosphorylation of acetate proceeds with the phosphorylation of the enzyme itself (36), as a phosphoenzyme can be isolated, and the isolated phosphoenzyme is able to transfer the phosphoryl group to either of the normal substrates, ADP and acetate (1). On the other hand, it is suggested that the phosphorylation of acetate proceeds with the direct in-line nucleophilic attack of the carboxylate O atom of acetate on the γ-phosphate of ATP (26), as the isotopic labeling data indicated that the phosphorylation of acetate proceeds with the inversion of the configuration of the γ-phosphate of ATP (3). Recent acetate kinase structures also support a direct in-line mechanism (17). The data on inversion of configuration can also be interpreted by the presence of two intermediates in the reaction, giving an odd number of transfers (36). The possibility of coexistence of the two mechanisms may explain the apparently conflicting data. Besides the mechanism, the substrate specificity is also worthy of investigation. The ability of Buk2, acetate kinase, and propionate kinase to differentiate between such similar substrates, which are different only in the number of methylene groups, will increase our understanding of enzyme specificity. To address these issues, the similarities to and differences from related proteins, the key residues in phosphoryl transfer, the substrate specificity, and the mechanism, we present the first structure of an enzyme from the butyrate/isobutyrate kinase family. The results lead to insights into the substrate specificity differences between Buk2 and acetate and propionate kinases and the reaction mechanisms of the three enzymes.

MATERIALS AND METHODS

Protein expression, purification, and crystallization.

While the purification was similar to that of regular hexahistidine-tagged Buk2 (13), the expression of His-tagged selenomethionine-substituted (SeMet) Buk2 as described below was different from that of regular His-tagged Buk2. The plasmid pET-30a(+)-buk2 was transformed into B834 (Met⁻) cells (Novagen). The cells were grown at 37°C in the LeMaster culture (28) containing the Kao-Michayluk vitamin solution (K3129; Sigma) and 50 mg/liter selenomethionine and were induced at an optical density at 600 nm of 0.8 to 1.1 with 1 mM isopropyl-β-d-thiogalactoside. After an additional 11 h of growth, the cells were harvested. The crystallization of His-tagged SeMet Buk2 was similar to that of regular His-tagged Buk2 (13). The protein solution contained 25 to 30 mg/ml His-tagged SeMet Buk2, 0.55 mM n-octyl-β-d-glucoside, 25 mM Tris-HCl (pH 8.5), 150 mM NaCl, 5 mM dithiothreitol, 10% (wt/vol) glycerol, 1 mM ADP, 1 mM MgCl₂, and 1 mM isobutyrate. The reservoir solution was initially 1 ml of 1.7 to 3.0 M sodium formate. The protein solution (2 μl or 4 μl) was mixed with an equal volume of the reservoir solution, and 0.1 to 0.2 ml of 1 M acetic acid (pH 4.5) was added only to the wells. Using the sitting-drop vapor diffusion method, crystals appeared in several days. The His-tagged SeMet Buk2 crystals grew more readily than those of regular His-tagged Buk2 and had better diffracting quality with stronger intensity and lower mosaicity. Adding 20 mM (β,γ-methylene) adenosine 5′-triphosphate (AMPPCP) to the sitting drops containing crystals produced AMPPCP-bound crystals. After being dipped in 4 M HCOOH (pH 4.5)-5% glycerol, the crystals were flash-cooled for data collection.

Data collection and processing.

SeMet multiwavelength anomalous dispersion data were collected to 3.0-Å resolution at wavelengths of 0.9799 Å, 0.9796 Å, and 0.9574 Å at the Advanced Photon Source BioCARS beamlines of Argonne National Laboratory (Table 1). At each wavelength, two sweeps of 90° data separated from each other by 180° were collected with an oscillation width of 1° and at a distance of 200 mm. Three sets of native data were also collected. One set of 180° of native data for regular His-tagged Buk2 was collected to 2.8-Å resolution at a wavelength of 1.0 Å. Another native set of 180° of His-tagged SeMet Buk2 data was collected to 2.9-Å resolution at a wavelength of 1.1 Å. The third native set of 360° of data for His-tagged SeMet Buk2 was collected to 2.5-Å resolution at a wavelength of 0.9 Å and was used in the final refinement (Table 2). Denzo and Scalepack (33) were used to process all the diffraction data. The crystals belonged to space group I422, and the cell dimensions of the His-tagged SeMet Buk2 crystal were 197.69 Å, 197.69 Å, and 58.24 Å.

TABLE 1.

Multiwavelength anomalous dispersion data collection^a

Parameter	Valueb
	Inflection	Peak	Remote
Wavelength (Å)	0.9799	0.9796	0.9574
Resolution (Å)	100-3.0 (3.11-3.0)	100-3.0 (3.11-3.0)	100-3.0 (3.11-3.0)
Rsym (%)	8.5 (50.1)	9.9 (53.5)	11.7 (78.5)
I/σ I	21.5 (3.1)	19.9 (2.6)	15.2 (1.7)
Completeness (%)	93.0 (81.4)	91.5 (78.7)	93.0 (82.4)
Redundancy	8.2 (4.0)	7.8 (3.9)	8.2 (4.3)
Figure of merit	0.59 (Solve)	0.74 (Resolve)

^aOne crystal was used to collect the MAD data, and another crystal was used to collect the native data for the final structural refinement (Table 2). Space group, I422; cell dimensions a, b, and c, 197.55, 197.55, and 57.73 Å; angles α, β, and γ, 90°.

^bValues in parentheses are for the highest-resolution shell.

TABLE 2.

Refinement statistics with native data^a

Parameter	Valueb
Space group	I422
Cell dimensions
a, c (Å)	197.69, 58.24
α, β, γ (°)	90
Wavelength (Å)	0.9
Resolution (Å)	100-2.5 (2.59-2.5)
No. of reflections	17,326 (1176)
Rsym (%)	6.1 (56.7)
I/σ I	15.9 (2.67)
Completeness (%)	85.4 (59.2)
Redundancy	11.9 (7.1)
Rwork/Rfree (%)	21.7/26.7
No. of atoms
Protein	2,959
Ligand/ion	44
Water	63
B factor (Å2)
Protein	46.51
Ligand/ion	67.16
Water	43.24
RMSD
Bond lengths (Å)	0.006
Bond angles (°)	1.3

^aOne crystal was used to collect the MAD data (Table 1), and another crystal was used to collect the native data for the final structural refinement.

^bValues in parentheses are for the highest-resolution shell.

Structure determination and refinement.

The fluorescence scan data from the His-tagged SeMet Buk2 crystal were measured at the Advanced Photon Source BioCARS beamlines. The anomalous scattering factors f′ and f″ were obtained with Chooch (15) and were input into Solve (38). Using the data between 40 and 3.75 Å, 11 out of 12 seleniums were found, and the initial phases were calculated using Solve and Resolve (37). The 3.75-Å experimental map was traceable, and a Cα trace was built into the electron density map using the program O (24). The Cα trace was superposed with the Cα trace of acetate kinase (Protein Data Bank [PDB] code 1G99), and the aligned secondary structures of acetate kinase were moved into the electron density map. The model was then used to generate a polyalanine chain, from which a mask was made for multicrystal averaging. After multicrystal averaging with the 2.8-Å regular His-tagged Buk2 data using DMMULTI (9) from CCP4 (7), the 2.8-Å map was much better. A partial model was built based on the sequence of Buk2 and the SeMet positions. After 10% of the reflections were selected as an R_free data set, the simulated annealing and conventional positional refinement using CNS (4) were alternated with the manual building in O. The updated model was used to generate a new mask for multicrystal averaging. The new mask making, multicrystal averaging, model building, and refinement were cycled until the model contained all 375 amino acids of Buk2. The 2.9-Å His-tagged SeMet Buk2 data were used in the later refinement, as the data brought the R factor down much more than did the 2.8-Å regular His-tagged Buk2 data. Using another set of 2.5-Å His-tagged SeMet Buk2 data, the final model was refined and obtained after B-factor refinement and the addition of 63 water molecules. The R factor and R_free dropped to 21.7% and 26.7%, respectively, between 43.6 and 2.5 Å. The same R_free data were used throughout all the data sets. All 375 amino acids from the Buk2 sequence were in the model, but the C-terminal six-histidine tag was not visible. The residues in the most favored, additional allowed, and generously allowed regions in the Ramachandran plot are 85.6%, 13.8%, and 0.6%, respectively. Figure 1 was produced with O, Fig. 2 with Powerpoint, and Fig. 6b with Chemdraw. Figures 3, 4, 5, 6a, 7, 8, and 9 were produced with Molscript (27) and Raster3d (30).

FIG. 1.

Overview of the final 2.5-Å 2fo-fc electron density map contoured at 1.0 σ level and fitted with the final model around residue Val 180.

FIG. 2.

Ribbon representation of the structure of Buk2. The blue color of the N terminus changes gradually to the red color of the C terminus. The N- and C-terminal domains are labeled N and C, respectively. AMPPCP is rendered in ball-and-stick mode, and the formate molecule is rendered in CPK mode.

FIG. 6.

(a) AMPPCP and 2fo-fc map contoured at 1.0 σ level. (b) Schematic diagram of AMPPCP binding to Buk2. Shown are all the residues hydrogen bonded to AMPPCP within 3.5 Å.

FIG. 3.

Topology diagram of Buk2. The secondary structures conserved in the ASKHA superfamily are rendered gray, and the inserts are shown in white. Rectangles indicate helices; arrows indicate β strands.

FIG. 4.

Stereo view of the active site of Buk2. Selected residues are shown as balls and sticks, as are formate and AMPPCP, and are colored according to atom type, where oxygen atoms are red, nitrogen atoms are blue, phosphorus atoms are purple, and carbon atoms are black.

FIG. 5.

Stereo view of the AMPPCP binding region. The four elements of secondary structure around AMPPCP are helix αe (Asp 255 to Gln 265), helix α2A′ (Gly 303 to Ala 306), a β turn (Mse 183 to Ile 187), and a U-shaped turn (Thr 211 to Arg 214). AMPPCP is rendered in ball-and-stick mode.

FIG. 7.

Stereo view of the probable isobutyrate binding site. The C-terminal domains of Buk2 and glycerol kinase are superimposed. Shown are the residues of Buk2 within a cover radius of 8 Å from glycerol. Only glycerol (red), phosphate (orange), and ADP (blue) from the glycerol kinase structure are included in the representation.

FIG. 8.

Octamer of Buk2. Each monomer is rendered in a different color, and AMPPCP is represented in ball-and-stick mode. The octamer can be viewed as being composed of four dimers, and the dimer is similar to that of acetate kinase. For example, the green and red monomers form a bird-like dimer. The eight cysteines in the octamer and the corresponding four disulfide bonds are represented in CPK mode.

FIG. 9.

Superposition of the Cα atoms of Buk2, acetate kinase (monomer A, PDB code 1G99), and propionate kinase (PDB code 1X3M). Buk2 is in green, acetate kinase is in both blue and red, and propionate kinase is in both cyan and purple. The N- and C-terminal domains are labeled N and C, respectively. AMPPCP is represented in ball-and-stick mode. The N-terminal β sheets from acetate and propionate kinases are not well superposed, as there are different relative rotations between the N- and C-terminal domains of acetate and propionate kinases.

Protein structure accession number.

The atomic coordinates of Buk2 from T. maritima have been deposited in the Protein Data Bank with the accession code 1saz.

RESULTS

Overall structure.

There is one Buk2 monomer in an asymmetric unit. The final 2.5-Å 2fo-fc map shows clear electron density for all 375 residues of Buk2 (Fig. 1), except that residues 228 to 241 from a loop of two helices have weak electron density. Buk2 folds like an open-shelled clam, with each of the two domains representing one of the two shells. The overall structure of Buk2 consists of two domains, the N-terminal domain (residues 1 to 130 and 332 to 375) and the C-terminal domain (residues 131 to 331) (Fig. 2 and 3). The residues of the C terminus (332 to 375) form two helices which attach to the N-terminal domain. Five β strands from the N-terminal domain form a β sheet, surrounded by five helices (α1, α2, αa, α3′, and αf). Similarly, five β strands from the C-terminal domain form another β sheet, surrounded by four helices (α1′, α2′, α3, and αe) and connected to the loop of two helices (αb and αc) (residues 220 to 242). The β sheets from the N- and C-terminal domains face each other, with an active-site cleft in between.

Active-site cleft.

Buk2 exhibits an open active-site cleft. In the active-site cleft, an AMPPCP molecule is bound to the C-terminal domain, a formate molecule is bound to the N-terminal domain, and a conserved residue, His 154, is located between AMPPCP and the formate molecule (Fig. 4). The distance between the γ-phosphate of AMPPCP and the formate molecule is about 9 Å, and the γ-phosphate of AMPPCP is hydrogen bonded to the imidazole ring of His 154 at a distance of 3.0 Å. A cluster of basic residues, Lys 15, Arg 74, His 104, His 154, His 182, Arg 214, and the neutral residue Asn 8, conserved among butyrate and acetate kinases, stretch into the active-site cleft (Fig. 4). The cleft may be closed during catalysis, and the conserved basic residues Arg 74, His 104, His 154, His 182, and Arg 214 may bind to the carboxyl group of isobutyrate or the ATP phosphate groups. The acidic residue Glu 334 in the cleft, which is invariant among butyrate kinases, is hydrogen bonded to the formate molecule. A cluster of hydrophobic residues, Leu 77, Val 130, Val 131, Ile 152, Ala 203, and Leu 204, down in the cleft may approach each other to bind to the hydrophobic part of isobutyrate, as the active-site cleft closes during catalysis.

AMPPCP binding region.

The adenine and ribose rings of AMPPCP are anchored in a groove of the C-terminal domain, and the phosphates hang over the inner wall of the cleft formed by the C-terminal domain, especially over the β turn (Mse 183 to Ile 187) (Fig. 5 and 6). The groove is formed by four elements of secondary structure: helix αe (Asp 255 to Gln 265), helix α2A′ (Gly 303 to Ala 306), the β turn (Mse 183 to Ile 187), and a U-shaped turn (Thr 211 to Arg 214). Most of these AMPPCP binding residues from the four elements of secondary structure are conserved among butyrate kinases. Abundant glycine residues are around AMPPCP, which is common in the nucleotide binding proteins, and some of the glycine residues interact with AMPPCP, such as Gly 185, Gly 186, and Gly 304. The adenine of AMPPCP is sandwiched between helices αe and α2A′, particularly between residues His 266 and Arg 257. A symmetry-related residue, Glu 22, participates in binding to adenine of AMPPCP (Fig. 6b), which suggests that Glu 22 may be the key to the recognition of ATP versus other nucleotides (GTP, TTP, and CTP). The ribose ring of AMPPCP is located at the meeting place of the four elements of secondary structure: the helices αe and α2A′, the β turn, and the U-shaped turn. The α,β-phosphates of AMPPCP are near the β turn (Mse 183 to Ile187), and the γ-phosphate is hydrogen bonded to His 154, Gly 185, Gly 186, and Arg 214 (Fig. 6b).

Formate binding site.

A V-shaped electron density probably indicates that a formate molecule is located on the side of the imidazole ring of His 154 and is surrounded by residues Arg 74, Glu 334, Pro 129, and Val 130 from the N-terminal domain. The sequence alignment of butyrate kinases indicates that the formate molecule is surrounded by conserved residues. The backbones of the conserved residues Arg 74-Gly 75-Gly 76, Val 127-Asp 128-Pro 129-Val 130 and His 154-Ala 155-Leu 156-Asn 157 are located approximately in parallel. A helix containing the conserved Glu 334 hangs over the above-mentioned three fragments of conserved residues. The conserved residues Asn 8, Lys 15, His 104, and Ser 106 are also near the formate molecule. The formate molecule is hydrogen bonded to the main chain carbonyl group of Asp128, the main-chain amide nitrogen group of Leu 156, and the side chains of Arg 74 and Glu 334 and is in a pocket formed by the side chains of Arg 74, His 154, and Glu 334, with Pro 129 and Val 130 underneath. This pocket may be formed owing to the opening of the active-site cleft. The formate binding site may or may not be one of binding sites of the isobutyrate, considering that the formate binding site and the probable isobutyrate binding site, as described below, will merge as the active-site cleft closes. Both the formate binding site and the probable isobutyrate binding site are around His 154; the former is on the side of the imidazole ring of His 154, while the latter is underneath the imidazole ring of His 154. The probable isobutyrate binding site is at a similar position of the acetate binding site as suggested in the acetate kinase structure (5).

Probable isobutyrate binding site.

There is no electron density for isobutyrate, although the crystallization buffer contains 1 mM isobutyrate. The superposition of the C-terminal domains of Buk2 and glycerol kinase containing glycerol (PDB code 1GlC) (16) suggests a binding site for isobutyrate (Fig. 7). Glycerol kinase is another member of the ASKHA superfamily of phosphotransferases. The probable binding site for isobutyrate is under the imidazole ring of His 154 and the γ-phosphate of AMPPCP, is over a cluster of hydrophobic residues (Leu 77, Val 130, Ile 152, Phe 153, and Leu 204), and is further surrounded by residues Val 130 to Val 132, Lys 150 to Leu 155, His 182, Gly 185 to Ile 189, Asn 201 to Asp 207, Glu 213, and Arg 214. Residues His 104, Ser 106, and Arg 74 from the N-terminal domain may also participate in binding to isobutyrate as the cleft closes during catalysis.

Octamer formation and interaction.

One Buk2 monomer exists in an asymmetric unit, and Buk2 packs as a donut-like octamer in the crystal (Fig. 8). The octamer is at a crystallographically special position of space group I422, with one octamer at (0, 0, z/2) and the other at (a/2, b/2, 0). The octamer itself has point group symmetry 422. In the octamer, two Buk2 monomers form a bird-like dimer through the interactions of the C-terminal domains, and four dimers related by a fourfold rotation form an octamer. The octamer has approximate dimensions of 140 Å by 140 Å by 60 Å. The previous data from dynamic light-scattering experiments predicted that the purified Buk2 is an octamer (13), and the data from analytical ultracentrifuge experiments also indicated that the purified Buk2 exists as an octamer (data not shown). Thus, both the dynamic light-scattering and the analytical ultracentrifuge experiments suggested that the oligomeric state of the active enzyme is an octamer. Interestingly, the four pairs of symmetry-related Cys 227 residues in the octamer form four intermolecular covalent sulfur-sulfur bonds in the crystal. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis without the reducing reagent showed that the purified Buk2 corresponds to a molecular mass of one monomer, suggesting that Buk2 is not disulfide linked before crystallization. The disulfide bond formation may be the result of oxidation of the crystal owing to long-term storage in the crystallization tray or exposure to synchrotron X-ray radiation. The close approach of symmetry-related pairs of cysteine 227 residues and the possible disulfide link may increase the thermal stability of Buk2 in thermophilic T. maritima.

The eight binding interfaces in the octamer are divided into two kinds of binding interfaces. One kind is formed by the two C-terminal domains of two monomers, and the other is through both the N- and C-terminal domains. The binding interface formed by the two C-terminal domains related by a twofold axis is similar to the ones in the crystals of acetate kinase and propionate kinase. This binding interface in Buk2 involves four regions: region A, residues Val 141 to His 144 and Gln 148 from the U-shaped secondary structure (positions 130 to 154); region B, residues Asp 199, Asp 205, Asp 207, Phe 210, Ser 215, Gly 246, and Tyr 250; region C, residues Leu 220 to Val 224, Cys 227, Phe 228, Phe 232, Tyr 234, and Mse 237 from the loop of two helices, αb and αc; and region D, residues Trp 269, Tyr 279, Gln 280, Lys 283, Lys 287, Ala 290, and Val 291 from the longest helix α1′ of the C-terminal domain and residue Arg 319 from helix α2B′. The protrusive A and C regions from each of the two symmetry-related monomers cross into each other, and the B and D regions from one monomer with a flat surface stack to another monomer. The other kind of binding interface involves fewer residues and consists of residues Glu 22, Mse 24, Val 25 to Asn 29, Mse 168, Asn 170, and Arg 257 only. Residues Glu 22, Mse 24, Mse 168, Asn 170, and Arg 257 have hydrogen-bonded and hydrophobic interactions with Mse 24, Mse 27, Mse 168, Val 328, Lys 309, Arg 347, and AMPPCP from another symmetry-related monomer. Residue Arg 23, though not involved in close contact with other residues, stretches the guanidine group into the cleft of another symmetry-related monomer, 6.2 Å and 5.4 Å away from the α and β phosphates of the symmetry-related AMPPCP, respectively, whereas the symmetry-related residue Arg 23 stretches back in a similar way. Two parts of Val 25 to Asn 29 from two symmetry-related β3 strands form an antiparallel β sheet. Similarly, in acetate kinase two symmetry-related βb strands from monomers A and B form a parallel β sheet (5).

Superposition of Buk2, acetate kinase, and propionate kinase.

Despite the low sequence identities of 24% and 19% between Buk2 and acetate kinase and propionate kinase, respectively, the overall folds of Buk2, acetate kinase (PDB code 1G99, 408 amino acids) (5), and propionate kinase (PDB codes 1X3M and 1X3N, 415 amino acids) (35) are quite similar to each other (Fig. 3 and 9). Buk2 and acetate and propionate kinases possess a similar ATP binding motif, which is composed of four elements of secondary structure: two helices, a β turn, and a U-shaped turn. Buk2 has three major folding differences from acetate and propionate kinases that share a same overall fold. Compared to acetate and propionate kinases, Buk2 lacks a two-stranded antiparallel β sheet beside strand β3, lacks a two-stranded antiparallel β sheet-like fragment beside strand β5′, and has an additional C-terminal helix αf (Fig. 9). The active-site clefts of Buk2 and acetate kinase are open to different extents. The cleft of Buk2 is about 12° more open than that of monomer A of acetate kinase. The clefts of Buk2 and monomer B of acetate kinase are open to the same extent at one end of the cleft, but at the other end of the cleft, the cleft of monomer B of acetate kinase is about 18° more open. The cleft of monomer A of acetate kinase is more closed than that of monomer B of acetate kinase, as a sulfate ion is bound at the active site of monomer A of acetate kinase. The active-site clefts of Buk2 and propionate kinase are open to a similar extent.

Using the program O, the superposition of the N-terminal domains between Buk2 and the two monomers of acetate kinase locates 151 (129) Cα atoms with a root mean square difference (RMSD) of 2.12 (1.98) Å, whereas the superposition of the C-terminal domains locates 223 (256) Cα atoms with an RMSD of 1.80 (1.90) Å; the values outside parentheses refer to the superposition between Buk2 and monomer A of acetate kinase, whereas those in parentheses refer to the superposition between Buk2 and monomer B of acetate kinase. The superposition between Buk2 and propionate kinase locates 291 (289) Cα atoms with an RMSD of 1.98 (1.96) Å; the values outside parentheses refer to the superposition between Buk2 and propionate kinase (PDB code 1X3M), whereas those in parentheses refer to the superposition between Buk2 and propionate kinase (PDB code 1X3N). Buk2, acetate kinase, and propionate kinase have common conserved residues in the active-site clefts. These residues may bind to the substrates and the phosphates of ATP and include Asn 8 (Asn 7 and Asn 11), Lys 15 (Lys 14 and Lys 18), Arg 74 (Arg 91 and Arg 86), His 104 (His 123 and His 118), His 154 (His 180 and His 175), His 182 (His 208 and His 203), Gly 184 (Gly 210 and Gly 205), Gly 186 (Gly 212 and Gly 207), Ser 188 (Ser 214 and Ser 208), Arg 214 (Arg 241 and Arg 236), Gly 304 (Gly 331 and Gly 326), Glu 334 (Glu 384 and Glu 381), and Glu 335 (Glu 385 and Glu 382); the residues outside parentheses refer to Buk2, whereas those in parentheses refer to acetate kinase and propionate kinase, respectively. Other common residues among Buk2, acetate kinase, and propionate kinase are located near adenine of AMPPCP and include Asp 255 (Asp 283 and Asp 278) and Arg 257 (Arg 285 and Arg 280). The noticeable feature is that at the active sites of Buk2 and acetate and propionate kinases, the four conserved basic residues are conserved in spatial arrangement; these include Arg 74 (Arg 91 and Arg 86), His 104 (His 123 and His 118), His 154 (His 180 and His 175), Arg 214 (Arg 241 and Arg 236), which may bind to the carboxyl groups of the substrates and the phosphates of ATP. At the active sites, Buk2 and acetate and propionate kinases have a cluster of hydrophobic residues, Leu 77 (Val 93 and Ala 88), Gly 102 (Ala 120 and Ala 115), Val 130 (Asp 148 and Asp 143), Ile 152 (Phe 179 and Phe 174), Ala 203 (Met 228 and Met 223), and Leu 204 (Pro 232 and Pro 227), which may bind to the hydrophobic parts of isobutyrate, acetate, and propionate, respectively. In particular, while the polar Asp 148 of acetate kinase and Asp 143 of propionate kinase occupy a similar position as Val 130 of Buk2, they may play a different role. Asp 148 and Asp 143 may provide a less hydrophobic active-site environment and further participate in binding to the polar parts of acetate and propionate, respectively, considering that Asp 148 and Asp 143 have longer side chains and that acetate and propionate are smaller substrates than isobutyrate.

Superposition of Buk2 and glycerol kinase.

Glycerol kinase is a member of the ASKHA superfamily of phosphotransferases, as it contains a core fold that is topologically similar to those of the members of the ASKHA superfamily (Fig. 3). Using the program O, the superposition of Buk2 and glycerol kinase (PDB code 1GLC) (16) shows that more than 55% of the Cα atoms of Buk2 have their counterparts in glycerol kinase. The superposition of the N-terminal domains between Buk2 and glycerol kinase locates 109 Cα atoms with an RMSD of 2.07 Å, and the superposition of the C-terminal domains locates 137 Cα atoms with an RMSD of 2.26 Å. Buk2 and glycerol kinase share a similar ATP binding motif: two helices, a β turn, and a U-shaped turn. Spatially, the cleft of Buk2 between the N- and C-terminal domains is largely open, whereas glycerol kinase (PDB code 1GLC) is a cleft-closed structure owing to the substrate binding at the active site. While Buk2 and glycerol kinase share core secondary structures, Buk2 has fewer inserts than glycerol kinase, considering that Buk2 has 375 amino acids and glycerol kinase has 501 amino acids. Between strands β4 and β5 (Fig. 3), Buk2 has a 49-residue insert which contains two helices and one strand, whereas glycerol kinase has a 156-residue insert which contains seven helices and six strands. One helix conserved in the ASKHA superfamily is also between strands β4 and β5. Between strands β2′ and β3′, glycerol kinase has a 22-residue insert of a two-stranded β sheet. While glycerol kinase has a counterpart helix of αe of Buk2 (Fig. 3), glycerol kinase has another 38-residue insert of a two-stranded β sheet between helices αe and α1′. Buk2 has a short 27-residue C terminus after Gly 348, whereas glycerol kinase has a long 44-residue C terminus. Compared to glycerol kinase, Buk2 has two additional inserts. One insert is between Val 130 and His 154 of Buk2, which is a U-shaped secondary structure element containing two small 3₁₀ helices, and the other insert is between Pro 212 and Gly 246 of Buk2, which is a loop of two helices. In glycerol kinase these two inserts are missing, and the residues in glycerol kinase corresponding to Val 130 and His 154, Pro 212 and Gly 246 of Buk2, respectively, connect directly. A number of residues around the active sites are conserved among Buk2, acetate kinase, and propionate kinase, but few are conserved between Buk2 and glycerol kinase or within the ASKHA superfamily.

DISCUSSION

Octamer formation.

Buk2 forms a donut-like octamer, composed of four bird-like dimers, whereas both acetate kinase and propionate kinase can form only a bird-like dimer. Buk2 has an overall fold different from that shared by acetate kinase and propionate kinase, and structural differences may result in different oligomeric states. In the Buk2 octamer there are two kinds of binding interfaces between monomers. One kind of binding interface is formed by the two C-terminal domains of two monomers, and this kind of binding interface is conserved in Buk2, acetate kinase, and propionate kinase. As a result, Buk2, acetate kinase, and propionate kinase all can form a bird-like dimer (Fig. 8). The other kind of binding interface is involved in strand β3 of the N-terminal domain and strand β5′ of the C-terminal domain, which are in contact with the neighboring Buk2 monomer. Compared to Buk2, both acetate kinase and propionate kinase have additional secondary structure elements beyond strands β3 and β5′ (Fig. 3 and 9), which would have clashes with the neighboring Buk2 monomer should acetate kinase or propionate kinase replace one Buk2 molecule in the Buk2 octamer. Hence, acetate and propionate kinases may form only a bird-like dimer, whereas Buk2 can form a donut-like octamer composed of four bird-like dimers.

Residues important in catalysis.

The superposition of Buk2, acetate kinase, and propionate kinase identifies the residues important in catalyzing phosphoryl transfer. At the active sites the four conserved basic residues are Arg 74 (Arg 91 and Arg 86), His 104 (His 123 and His 118), His 154 (His 180 and His 175), and Arg 214 (Arg 241 and Arg 236), which may bind to the carboxyl groups of the substrates and the phosphates of ATP; the residues outside parentheses refer to Buk2, whereas those in parentheses refer to acetate and propionate kinases, respectively. At the active sites there is a cluster of hydrophobic residues, Leu 77 (Val 93 and Ala 88), Gly 102 (Ala 120 and Ala 115), Val 130 (Asp 148 and Asp 143), Ile 152 (Phe 179 and Phe 174), Ala 203 (Met 228 and Met 223), and Leu 204 (Pro 232 and Pro 227), which may bind to the hydrophobic parts of isobutyrate, acetate, and propionate, respectively. In particular, as described in “Superposition of Buk2, acetate kinase, and propionate kinase” above, Asp 148 of acetate kinase and Asp 143 of propionate kinase may participate in binding to the polar parts of acetate and propionate. The four basic residues are conserved among the three enzyme families, whereas the hydrophobic residues at the active sites are not well conserved even in each of the three enzyme families. During evolution, butyrate, acetate, and propionate kinases put high restraints on the conservation of these four basic residues. It is likely that these four basic residues are required to bind to the conserved carboxyl groups of the substrates or the phosphates of ATP and bring the substrates close to ATP for phosphorylation. In contrast, butyrate, acetate, and propionate kinases put loose restraints on the residues interacting with the hydrophobic parts of the substrates, so that the hydrophobic residues have a diverse evolution and are not well conserved in each of the three enzyme families.

Substrate specificity differences.

Comparison of the active-site residues, especially the hydrophobic ones, identifies the substrate specificity differences among Buk2 and acetate and propionate kinases. Acetate and propionate kinases have almost the same active-site hydrophobic residues, consistent with the sharing of a common overall fold between these two enzymes. The sole difference of Ala 88 in propionate kinase and Val 93 in acetate kinase ensures a larger substrate binding cavity in propionate kinase (35). There are four aspects of substrate specificity differences between Buk2 and acetate kinase and propionate kinase. First, compared to Ala 120, Phe 179, and Met 228 in acetate kinase and Ala 115, Phe 174, and Met 223 in propionate kinase, the smaller Gly 102, Ile 152, and Ala 203 in Buk2 ensure a larger substrate binding cavity. Second, the opposing residues Leu 77 and Leu 204 from the N- and C-terminal domains of Buk2 have longer side chains than their counterparts Val 93 and Pro 232 in acetate kinase and Ala 88 and Pro 227 in propionate kinase, respectively, and may contribute to a larger cavity when the active-site cleft is closed. Third, compared to the three-residue fragments Arg 91-Val 92-Val 93 in acetate kinase and Arg 86-Ile 87-Ala 88 in propionate kinase, the four-residue fragment Arg 74-Gly 75-Gly 76-Leu 77 in Buk2 contributes to a larger substrate binding cavity. The residues Arg 74 and Leu 77 in Buk2 (Arg 91 and Val 93 in acetate kinase and Arg 86 and Ala 88 in propionate kinase) may interact with the polar and hydrophobic parts of the substrates, respectively, and the residues Gly 75 and Gly 76 without side chains would leave more space to form a larger cavity. The side chain of Val 92 in acetate kinase and the side chain of Ile 87 in propionate kinase point away from the active sites, in a direction opposite to those of Arg 91 and Val 93 in acetate kinase and those of Arg 86 and Ala 88 in propionate kinase, respectively, and do not contribute to the formation of the substrate binding cavities. Fourth, compared to Asp 148 in acetate kinase and Asp 143 in propionate kinase, Val 130 in Buk2 makes the active site more hydrophobic, which is consistent with the different hydrophobicity among the substrates. As isobutyrate is more hydrophobic than acetate and propionate, isobutyrate may need a more hydrophobic active-site environment.

Enzyme mechanisms.

Buk2, acetate kinase, and propionate kinase share a similar fold, a similar ATP binding motif, and some conserved residues around the active sites, suggesting that they are likely to share a similar mechanism. The biochemical data for acetate kinase showed the inversion of configuration of the γ-phosphate of ATP during the phosphorylation of acetate (3, 26). Thus, Buk2, acetate kinase, and propionate kinase may have the three following reaction mechanisms, except that in acetate and propionate kinases the active-site residues may be substituted accordingly. The first is a direct in-line mechanism (17) as is thought to occur in glycerol kinase and other members of the ASKHA superfamily of phosphotransferases. Descending from the same ancestor, it seems likely that the superfamily members follow a common mechanism (19). The four basic residues Arg 74, His 104, His 154, and Arg 214 in Buk2 may bind to the substrate so that the active-site cleft closes. After substrate binding, the four basic residues may also bind to the phosphates of ATP, resulting in tension and easier breakage of the bond between the gamma and beta phosphates. In addition, by binding to the phosphates, the positive charges of the four basic residues can be neutralized by the phosphates, so that the carboxylate O atom of the substrate is not withdrawn too much by the four basic residues and the nucleophility of the carboxylate O atom remains. The following attack on the γ-phosphate by the carboxylate O atom of the substrate results in the inversion of configuration of the γ-phosphate. The second is an indirect mechanism. Because the γ-phosphate of AMPPCP is hydrogen bonded to the imidazole ring of the conserved His 154 in Buk2 (Fig. 4 and 6b), ATP may phosphorylate His 154 with the γ-phosphate and the phosphoryl group may then be transferred to another intermediate residue before phosphorylating isobutyrate. In this way, two phosphorylated intermediates are involved and three SN2 nucleophilic displacement reactions occur in the catalysis (36), which ensures a net inversion of configuration of the γ-phosphate. The third possibility is also an indirect mechanism. ATP phosphorylates the imidazole ring of His 154 in Buk2 with the inversion of configuration of the γ-phosphate, followed by an adjacent attack of the carboxylate O atom of isobutyrate on the bound phosphoryl group. Adjacent attack is one common nucleophilic displacement mechanism in organophosphorous stereochemistry (10), resulting in the retention of configuration of the phosphoryl group in the reaction. In this way, a net inversion of configuration of the γ-phosphate also occurs in the catalysis. In addition, imidazole can act as a catalyst in organophosphorous phosphorylation reactions (34). In nucleoside diphosphate kinase (31), for example, the imidazole ring of a histidine acts as a phosphorylated enzyme intermediate. That the activity of acetate kinase decreases dramatically after residue His 180, the counterpart of His 154 in Buk2, is mutated to Ala 180 (23) suggests the importance of His 154 in Buk2 and a possibility of an indirect reaction mechanism.