The FASEB Journal • Research Communication
Molecular architecture and structural basis of allosteric regulation of eukaryotic phosphofructokinases Norbert Stra¨ter,*,1 Sascha Marek,* E. Bartholomeus Kuettner,* Marco Kloos,* Antje Keim,* Antje Bru¨ser,† Ju¨rgen Kirchberger,† and Torsten Scho¨neberg†,1 *Institute of Bioanalytical Chemistry, Center for Biotechnology and Biomedicine, and †Institute of Biochemistry, Medical Faculty, University of Leipzig, Leipzig, Germany Eukaryotic ATP-dependent 6-phosphofructokinases (Pfks) differ from their bacterial counterparts in a much more complex structural organization and allosteric regulation. Pichia pastoris Pfk (PpPfk) is, with ⬃1 MDa, the most complex and probably largest eukaryotic Pfk. We have determined the crystal structure of full-length PpPfk to 3.05 Å resolution in the T state. PpPfk forms a (␣␥)4 dodecamer of D2 symmetry with dimensions of 161 ⴛ 157 ⴛ 233 Å mainly via interactions of the ␣ chains. The N-terminal domains of the ␣ and  chains have folds that are distantly related to glyoxalase I, but the active sites are no longer functional. Interestingly, these domains located at the 2 distal ends of this protein along the long 2-fold axis form a (␣)2 dimer as does the core Pfk domains; however, the domains are swapped across the tetramerization interface. In PpPfk, the unique ␥ subunit participates in oligomerization of the ␣ chains. This modulator protein was acquired from an ancient S-adenosylmethionine-dependent methyltransferase. The identification of novel ATP binding sites, which do not correspond to the bacterial catalytic or effector binding sites, point to marked structural and functional differences between bacterial and eukaryotic Pfks.—Stra¨ter, N., Marek, S., Kuettner, E. B., Kloos, M., Keim, A., Bru¨ser, A., Kirchberger, J., Scho¨neberg, T. Molecular architecture and structural basis of allosteric regulation of eukaryotic phosphofructokinases. FASEB J. 25, 89 –98 (2011). www.fasebj.org
ABSTRACT
Key Words: metabolism 䡠 protein crystallography 䡠 structural biology
The phosphorylation of fructose 6-phosphate (F-6-P) to fructose 1,6-bisphosphate is a key control step of glycolysis in most organisms. It is catalyzed by the ATP-dependent 6-phosphofructokinase (Pfk, EC 2.7.1.11), an enzyme that shows manifold allosteric regulation. Pfks from bacteria were among the first enzymes to which the Monod model of allosteric cooperativity was applied (1). Bacterial Pfks are modulated essentially by only 2 effector molecules: Mg2⫹/ADP acts as an activator, whereas phosphoenolpyruvate (PEP) is inhibitory. The structural basis of the catalytic steps and allosteric changes has been characterized early based 0892-6638/11/0025-0089 © FASEB
on crystal structures of Escherichia coli Pfk (EcPfk) and Bacillus stearothermophilus Pfk (BsPfk) (2– 6). More recently, structures of Pfks from Lactobacillus bulgaricus, Listeria innocua (unpublished observations; PDB 3HIC and 3IE7), and Trypanosoma brucei have also been determined (7–10). The latter structure represents the first eukaryotic Pfk structure; however, this homotetrameric enzyme rather resembles the bacterial diphosphate-dependent enzymes and does not possess the characteristic features of eukaryotic Pfks (9). Bacterial Pfks have a homotetrameric structure of D2 symmetry (6). Each monomer of ⬃320 aa contains 1 active site with the ATP and F-6-P binding sites and an effector binding site for PEP or ADP. The binding sites for F-6-P and the effectors are located at the interface between subunits. Eukaryotic Pfks exhibit a far more complex and sophisticated regulatory mechanism. They differ significantly in their allosteric regulation from the bacterial Pfks, being influenced by ⬎20 effectors (11). Physiologically most relevant is the activation by AMP and fructose 2,6-bisphosphate (Fru-2,6-P2), whereas ATP and citrate exert an inhibitory effect. Based on sequence homology, it has been suggested that the eukaryotic Pfks evolved from an ancestor protein resembling the current bacterial Pfks by a gene duplication and tandem fusion event (12, 13). Each subunit of a eukaryotic Pfk thus consists of 2 homologous halves, each related to a bacterial Pfk subunit. The better conserved N-terminal halves contain the active site (termed F and N in this study for the F-6-P and nucleotide binding sites, respectively), whereas the former catalytic site of the C-terminal half may have developed into new regulatory binding sites (F⬘ and N⬘). In the well-studied Saccharomyces cerevisae Pfk (ScPfk), a further gene duplication event and divergence resulted in 2 homologous subunits, forming ␣44-heterooctamers of D2 symmetry (14, 15). Both 1 Correspondence: N.S., Institute of Bioanalytical Chemistry, Center for Biotechnology and Biomedicine, University of Leipzig, Deutscher Platz 5, 04103 Leipzig. E-mail:
[email protected]; T.S., Institute of Biochemistry, Medical Faculty, University of Leipzig, Johannisallee 30, 04103 Leipzig. E-mail:
[email protected] doi: 10.1096/fj.10-163865 This article includes supplemental data. Please visit http:// www.fasebj.org to obtain this information.
89
chains of yeast Pfks also contain a ⬃200 residue Nterminal domain of unknown function that has no significant sequence homology to structurally or functionally characterized domains. In mammals, these N-terminal domains are lacking. Although dimeric forms of mammalian Pfks were detected, the smallest active form is the tetramer of ⬃330 kDa (12). All mammalian enzymes show a tendency to form larger aggregates (16), and they can form isoforms in a tissue-dependent manner. In humans, Pfk is encoded by 3 genes (PFK-M, PFK-L, and PFK-P) that are located on different chromosomes. Despite early reports of the crystallization of rabbit muscle Pfk (17) and later of a proteolyzed fragment of ScPfk (18), the structure of the eukaryotic Pfks has remained elusive until now. Three-dimensional electron microscopy studies (19 –23) and small-angle X-ray scattering experiments (24) revealed low-resolution structures of the molecular assembly of the subunits of yeast Pfks. Pichia pastoris Pfk (PpPfk), with a molecular mass of 975 kDa, is one of the largest and more complex representatives of the Pfk family (25–27). It contains a unique ␥ subunit of 41 kDa that has so far only been characterized in Pichia species, and no protein homologue of known structure or function could be identified. The ␣ and  subunits of 109 and 104 kDa molecular mass, respectively, share high structural homology (54 – 60% amino acid identity) to the respective subunits in other heterooligomeric yeast Pfks, such as ScPfk. Although the unique ␥ subunit is not essential for enzymatic activity, it modulates the allosteric behavior of the enzyme and contributes to the adaption of P. pastoris to variation of the energy charge of the cell (27). The ␥ subunit is an integral part of the (␣␥)4-PpPfk, and based on an electron microscopy structure, it bridges the N-terminal halves of the ␣ and  subunits (26, 28). The spatial structure and catalytic mechanism of bacterial Pfks have been characterized now ⬎20 yr ago; however, the structure and mechanism of allosteric regulation of eukaryotic Pfks cannot be inferred from the bacterial Pfk structures. Unfortunately, many of the large eukaryotic Pfks tend to aggregate and could not be crystallized so far. To characterize the molecular basis of allosteric regulation of the complex eukaryotic Pfks, we determined the crystal structure of PpPfk. The complete untruncated 975-kDa enzyme crystallized in the presence of inhibiting concentrations of ATP. Thus, a model of the T state of PpPfk was refined at 3.05 Å resolution, providing the first fundamental insight into the molecular architecture of eukaryotic Pfks. The structure also reveals the evolutionary origins of the ␥ subunit and of the N-terminal domains of the ␣ and  subunits. Furthermore, the location of the ATP effector binding site was identified at a position not corresponding to any bacterial effector binding site, providing, together with comparisons with the bacterial Pfk structures, insight into the different regulatory mechanisms of bacterial and eukaryotic Pfks. 90
Vol. 25
January 2011
MATERIALS AND METHODS PpPfk was purified essentially as described previously (26, 27). After ammonium sulfate precipitation, the enzyme was stored overnight as a pellet in the presence of 1 mM ATP at ⫺18°C. For crystallization, the enzyme was concentrated to ⬃10 mg/ml in a buffer containing 10 mM imidazol, 5 mM mercaptoethanol, 10% glycerol, and 10 mM ATP, pH 7.0. The 1-l protein solution and 1-l crystallization buffer were equilibrated against the reservoir solution containing 0.7 M ammonium sulfate and 0.1 M sodium citrate, pH 4.6. Microseeding (streak seeding) was essential for the growth of larger crystals of ⬎100 m size. The crystals were stepwise (⬍1 min total soaking time) transferred to a buffer containing 1 M (NH4)2SO4, 20% PEG200, and 0.1 M sodium citrate, pH 4.6, and frozen in liquid nitrogen. In the course of this project, many crystals had to be tested at different synchrotron beamlines, since the crystal quality after cryocooling generally differed considerably. Furthermore, the crystals tend to break easily during the cryocooling step and split spots were often observed. Details of data collection and refinement are summarized in Table 1. The 232 frames were collected using a MX-225 (Rayonix LLC, Evanston, IL, USA) detector at beamline 14.2 at Berliner Elektronenspeicherring-Gesellschaft fu¨r Synchrotronstrahlung (BESSY II; Berlin, Germany) with an oscillation angle of 0.5° and an exposure time of 20 s. The detector was placed at a distance of 340 mm behind the sample. The phase problem was solved by molecular replacement trials (program MolRep) using a tetramer, dimer, or monomer of EcPfk (PDB 1PFK) as the search model. A clear solution matching the arrangement of the electron microscopy (EM) structures (19, 20, 28) was found by placing 4 tetramers of the bacterial enzyme in the asymmetric unit (r⫽47.1% after rigid body refinement). The electron density TABLE 1. Data collection and refinement statistics Statistic
Space group X-ray source Wavelength (Å) Temperature (K) Unit cell (Å) a b c  Resolution (Å) Mosaicity (deg) Completeness (%) Unique reflections Multiplicity Rsym (%) Rpim (%) Average I/(I ) Wilson B (Å2) Rwork/Rfree (%) RMSD bonds (Å)/angles (deg) Average B value (Å2) Chains A/B/I Chains C/D/J Chains E/F/K Chains G/H/L ATP: B/D/F/H Sulfate ions
Value
P21 BESSY BL 14.2 0.918 100 161.7 188.3 231.6 92.8 29.8–3.05 (3.21–3.05) 0.16–0.25 97.5 (98.5) 255840 2.6 (2.6) 11.3 (56.8) 8.6 (42.6) 8.3 (1.9) 75.8 20.0/23.2 0.010/1.24 68.2 58.4/59.6/51.0 59.6/59.4/50.6 78.5/86.5/67.6 73.5/79.8/75.5 41.6/44.8/90.7/77.2 95.5
Values in parentheses refer to the highest-resolution shell.
The FASEB Journal 䡠 www.fasebj.org
STRA¨TER ET AL.
map was significantly improved by phase refinement including 4-fold symmetry averaging (program DM; ref. 29) such that the ␣ and  chains of the enzyme could be distinguished at regions that are well conserved between the bacterial and the yeast Pfks. After modeling and refinement (program Refmac5; ref. 30) of the conserved domains of the ␣ and  chains, the electron density of the ␥ chain became apparent and a molecular model could be built. At this point, refinement was continued with Buster-TNT (31). Unmodeled density was identified at the distal ends of the molecule along the z axis. Even in the DM-improved density maps, the quality of these regions was significantly worse than the previously modeled parts including the ␥ subunits, which are largely very well defined in the electron density maps considering the moderate resolution of the data. For all initial model building steps, maps after phase improvement including NCS averaging were used. Noncrystallographic symmetry restraints and TLS refinement were employed in the refinement up to the final structure. Separate TLS groups were defined for residues 5–186, 208 –989 of the ␣ chain; residues 5–155, 180 –941 of the  chain; and residues 5–351 of the ␥ chain. The final model comprised residues 5– 42, 44 –54, 56 –74, 77–98, 100 – 102, 104 –114, 118 –152, 155–176, 178 –186, 208 –333, 342– 760, 762– 819, 823–962, 967–989 of the ␣ chain; residues 5–19, 21, 23– 45, 49 – 83, 94 –143, 145–155, 180 –305, 314 –363, 365–558, 560 –736, 738 – 897, 899 –920, 922–941 of the  chain; residues 5–151, 176 –351 of the ␥ chain; 4 ATP ions; and 71 sulfate ions. The structure was validated using the programs Procheck (32) and MolProbity (33). Of the residues, 89.2% were in the most favored regions, 10.5% were in the additionally allowed regions, 0.2% (19 residues) were in the generously allowed regions, and no residues were in the disallowed regions. Figures were prepared using Pymol (http://www.pymol.org) and Aline (http://crystal.bcs.uwa. edu.au/px/charlie/software/aline). The coordinates have been deposited in the Protein Data Bank (PDB 3opy).
RESULTS AND DISCUSSION Oligomer structure and relationship to bacterial Pfks The crystallographic model of PpPfk consists of 8596 amino acid residues of the (␣␥)4 heterooligomer in the asymmetric unit; 528 residues were not included in the model, since they were part of flexible regions and that could not be accurately modeled in the electron density maps. In addition, 4 ATP molecules and 71 sulfate ions were included in the model. The heterododecamer has D2 symmetry and dimensions of 161 ⫻ 157 ⫻ 233 Å along the 3 perpendicular 2-fold axes, which are referred to as x, y, and z (Figs. 1 and 2). The C-terminal halves of the ␣ subunits interact with each other along the x axis close to the center of the molecule. In contrast, large solvent channels are present along the y axis with almost no contacts between the ␣ chains. Along the long z axis, only the N-terminal domains of the ␣ subunits are in loose contact, otherwise a long solvent channel runs through the heterododecamer in this direction. The ␥ subunits bridge the ␣ and  chains close to the center of the molecule. The ␣ and  chains form a heterodimer that corresponds to a bacterial Pfk homotetramer such that the molecular symmetry axes denoted p, q, and r of the bacterial enzyme of D2 symmetry (6) are pseudosymmetry axes of the ␣ dimer in PpPfk (these pseudodyads relate the following regions in the ␣-heterodimer: p axis, N-terminal half with C-terminal half within each chain; q axis, N-terminal half of one chain with the
Figure 1. Structure of the PpPfk (␣␥)4-dodecamer. The ␣, , and ␥ chains are shown in blue, red, and yellow, respectively. ATP molecules bound to the  chains are shown as green spheres. Further ligands are superimposed from the E. coli structure (PDB 1PFK) to mark further putative binding sites: ADP (pink spheres) and F-6-P (white spheres) in the current and former active site and ADP (yellow sticks) in the bacterial effector binding sites. Loops shown in white in this and other figures are disordered and not part of the refined structure. A) View along the 2-fold axis along x. B) After a 40.5° rotation around the z axis, view is along the pseudodyad r of the ␣ dimer in the top half. Substrate models in the active sites of the neighboring N-terminal halves of the ␣ and  chains mark the catalytic sites. C) View along the y dyad. X-RAY STRUCTURE OF P. PASTORIS PHOSPHOFRUCTOKINASE
91
Figure 2. Scheme of the architecture and arrangement of the ␣ and  chains in the PpPfk dodecamer. The ␥ subunits are omitted for clarity. The ␣ chain is shown in purple (glyoxalase I-like domain), blue (N-terminal half), and light blue (Cterminal half). Corresponding regions of the  chain are in yellow, red, and orange. Residue numbers are indicated for the beginning and end of these regions. View corresponds to Figs. 1B and 2A. Pseudodyads r, p, and q relating the N- and C-terminal halves of the ␣ and  chains are indicated. x, y, and z are the molecular dyads relating the 4 equivalent ␣␥protomers. The x and y axes are located at the octamerization interface; z axis relates 2 ␣ heterotetramers at the tetramerization interface formed between the C-terminal halves of these chains. Second heterotetramer in the top half of the figure would lie behind it and is omitted for clarity.
C-terminal half of the other chain; and r axis, Nterminal half ␣ chain with N-terminal half  chain; see Fig. 2 and Supplemental Fig. 1). In agreement with recent mutational and kinetic studies (34, 35), the gene fusion resulted in an arrangement, where the N-terminal halves of both chains interact with each other at one side of the heterodimer and the C-terminal halves at the other side. Two ␣-heterodimers tetramerize via the 2-fold z axis by interactions between the C-terminal halves of the ␣ and  chains such that the r axes of the 2 ␣ dimers are almost colinear and thus perpendicular to the z axis. This interaction at the tetramerization interface (for comparison with other yeast or eukaryotic Pfks, we denote oligomerization interfaces as “tetramerization” or “octamerization” interfaces although 92
Vol. 25
January 2011
in PpPfk the ␣␥ protomer would assemble to a hexamer or dodecamer, respectively; these terms thus refer to ␣ chains only) is mainly mediated by helices 8⬘ and 9⬘ [see Supplemental Fig. 2; secondary structure elements are denoted such that the same identifier (numbers for helices, letters for strands) is used as for corresponding element in bacterial enzymes. A prime is added for elements in the C-terminal half. A complete list is given in sequence alignment; see Supplemental Fig. 14]. In the resulting heterotetramer, the N-terminal halves of the 2 chains with the active sites point to the outside, whereas the putative Fru-2,6-P2 allosteric sites (36) point to the central channel along the z axis. This ␣22-heterotetramer most likely corresponds to the mammalian homotetramers. In heterooligomeric yeast Pfks, 2 ␣22-tetramers form an ␣44-octamer by the action of dyads x and y. Interaction of the ␣ chains at the octamerization interface involves the loop between helix 5⬘ and strand E⬘; the long loop between helix 11⬘ and strand J⬘, including helix 11A⬘, which is not present in the bacterial enzymes; and helices 2⬘ and 4⬘. In ScPfk, proteolytic cleavage and removal of ⬃80 aa from the C terminus of the ␣ chain results in the formation of a catalytically active tetrameric 12S-Pfk (14). The corresponding interface formed by the  subunits at the distal ends of the molecule along the z axis is involved in interactions with the N-terminal domains of the ␣ chains. These domains might thus prevent further aggregation along the z axis. For the mammalian Pfks, which lack the N-terminal domains, the formation of higher oligomers and heteroassociation with other proteins have been observed (16, 37, 38). Catalytic site and effector binding sites Both active site structures in the N-terminal halves of the ␣ and  chains appear to be intact, and in each active site of the 4 ␣ chains, 3 sulfate ions are bound close to the positions of the ATP ␥-phosphate and both phosphate moieties of F-1,6-P2 (Supplemental Figs. 3A and 4). In the  chains, the density indicates that ATP and sulfate ions are bound with partial occupancy. However, due to the weak density, only the sulfate ions have been included in the refined model. These findings are in agreement with previous studies (12, 39, 40) that indicated that the N-terminal halves of both chains are catalytically active. The former active site structures of the C-terminal halves have changed significantly (Supplemental Fig. 3C, D). In particular, the nucleotide binding site appears to be completely destroyed, which strongly suggests that this is not the effector binding site for AMP or ATP. The sugar binding site is better maintained and may bind the activating F-2,6-P2 effector, as previously indicated by mutagenesis studies for ScPfk (36) and the mammalian enzymes (39, 41). The heterotropic allosteric coupling mechanism between the new effector binding site in the C-terminal halves and the conserved catalytic site in the N-terminal halves is probably similar to the allosteric mechanism of homotropic cooperativity as characterized for the bacterial enzymes, since the arrangement of subunits in
The FASEB Journal 䡠 www.fasebj.org
STRA¨TER ET AL.
Figure 3. Arrangement of the domains in the ␣ dimer. A) View along the pseudodyad r. The ␣ chain is shown in purple (glyoxalase I-like domain), blue (N-terminal half) and light blue (C-terminal half). Corresponding regions of the  chain are in yellow, red, and orange. The ␥ subunit is shown in pale yellow. ATP molecule and superposed ligands are shown as in Fig. 1. B) Potential effector binding sites of 1 ␣ dimer (without N-terminal domains). E, effector binding sites of the bacterial Pfks; N, nucleotide binding site in the bacterial active sites; F, F-6-P site of the bacterial active site. Prime indicates site that belongs to the C-terminal halves. As an example, N⬘ marks the former catalytic nucleotide binding site in the C-terminal half of the  chain.
the ␣-heterodimer resembles that of the bacterial homotetramer. The 4 equivalent effector binding sites of the bacterial homotetramer are located between the subunits. In PpPfk, these sites correspond to 4 distinct potential effector binding sites per ␣ dimer (in addition to further sites derived from the former catalytic sites, see Fig. 3B). Two sites (␣E and ␣E⬘, see Fig. 3 for notation) are located between the N- and C-terminal halves of the ␣ chain and 2 further sites (E and E⬘) at the corresponding positions in the  chain. Sites ␣E and E, mainly formed by residues of the region between helix 8 and strand H and by residues following strand C⬘, show significant differences compared with the bacterial enzymes, but it appears possible for both chains that these sites still function as effector binding sites. Mutational studies (39) on the mammalian enzymes indicated that these sites correspond to the ATP inhibitory binding site. Sites ␣E⬘ and E⬘, formed by the loop between helix 8⬘ and strand H⬘ and by residues following strand C, are more diverged in both chains compared with the bacterial enzymes, in particular residues 818 – 825 of the ␣ chain and the corresponding residues 792 to 799 of the  chain occupy the position of the nucleoside moiety of the ADP complex of the
bacterial Pfk (Supplemental Fig. 5). Polar residues, including 4 positively charged side chains, line the former phosphate binding pocket. These sites (corresponding to ␣E⬘ and E⬘) have been implicated in binding the inhibitory citrate in mammalian Pfks (39, 41), which appears plausible also for PpPfk based on the size and positive charge of the binding pocket. However, despite the presence of 100 mM citrate in the crystallization buffer, this binding site is occupied by sulfate ions. PpPfk is inhibited by 10 mM citrate in the absence of AMP to 20% residual activity (data not shown). Besides AMP, the presence of NH4⫹ also abolishes citrate inhibition (42). It is reasonable to assume that the high concentration of sulfate and ammonium ions used as a precipitant in crystallization replaces citrate from its binding site. Citrate molecules were also not identified at other binding sites. We find 4 clearly defined ATP molecules bound to the small domain of the N-terminal half of the  chain (Fig. 4). This ATP binding pocket is mainly formed by helices 8 and 13 and by strand G at the edge of the central  sheet of the small domain. The adenine base is oriented deep inside the pocket, whereas the triphosphate chain is positioned at the surface of the protein. When compared with the corresponding position in
Figure 4. ATP binding site in the  chains. A) View into the binding pocket. Electron density of ATP is shown in black (2Fo–Fc-type map). B) Scheme of interactions of ATP ligand and protein. X-RAY STRUCTURE OF P. PASTORIS PHOSPHOFRUCTOKINASE
93
the bacterial enzymes, the loop between strand G and helix 8, which coordinates the triphosphate chain, represents the most significant difference (Supplemental Fig. 6). Interestingly, there are further unique structural features of the eukaryotic enzymes nearby: helix 12 is much longer than in the bacterial Pfks, an additional helix 12A is formed, and there is a linker between the N-terminal half and the C-terminal half (after helix 13), which is also near the ATP binding site. Furthermore, the loop between strand H and helix 9 is longer than in the bacterial enzymes. These additional structural elements are conserved in other eukaryotic Pfks, although they are not involved in any subunit interactions. The ATP molecule is 11 Å away from the bacterial effector binding site, which is presumed to be the inhibitory ATP binding site (corresponding to ␣E and E) in mammalian Pfks based on mutagenesis data (39). Although a P728L mutation in the ␣ chain of ScPfk close to the F⬘ binding site has been shown to abolish ATP activation (43), the location of the inhibitory ATP binding site remains unknown, since the mutation of this buried proline might destroy the activation mechanism but not the binding site. Furthermore, this mutation also abolishes activation by AMP and F-2,6-P2. The following arguments support the notion that the ATP molecule complexed to the PpPfk structure is bound to the inhibitory ATP site. Sites ␣E and E are not occupied by ATP in our structure, and they are also not blocked by sulfate ions. ATP was present at a concentration of 10 mM in the crystallization buffer, and the presence of ATP is important for the stability of the enzyme. Finally, the enzyme was crystallized in the T state (see below) and the inhibitory ATP binding site should be occupied under these conditions. A comparison to bacterial enzymes suggests a possible mechanism as to how ATP binding to this site might influence the allosteric transition. Helices 8 and 9 of the bacterial Pfks perform a shear motion with a net translation of 1.2 Å on the central  sheet of the small domain between the R and T states (6). ATP binding might induce this shear motion by binding between helix 8 and strand H of the central  sheet. If helices 8 and 9 are shifted toward the active site of the ␣ chain as a result of ATP binding between helices 8 and 13, strands F and I might also shift closer to the ␣ chain, thereby inducing the closure of the F-6-P binding site similar to the situation in the bacterial enzymes, albeit induced by a different mechanism. Strands F and I form part of the active site, and they include residues ␣R396/R368 and ␣R487/R460, which are positioned to bind the 6-phosphate group of the substrate in PpPfk. A comparison of the ATP binding site of the  chain with the corresponding site of the ␣ chain shows that the adenine and ribose binding pockets appear well conserved, with no or only conservative replacements (Supplemental Fig. 6). The main difference is seen in the loop that binds the terminal phosphate group, which is oriented away from the binding site in the ␣ chain. In particular, the replacements of S404 and Q410 of the  chain by E432 and D438 in the ␣ chain 94
Vol. 25
January 2011
might negatively affect binding of ATP. There are no direct crystal packing interactions influencing the conformation of this loop. However, it is possible that crystal packing interactions indirectly interfere with a possible ATP binding site in the ␣ chain by stabilizing the allosteric protein in a conformation that does not allow binding of ATP to the ␣ chain. Allosteric state Crystallization in the presence of 10 mM ATP in the absence of activators and F-6-P suggests that the enzyme is in the T state (KTATP⫽0.024 mM; ref. 26). Based on a comparison of the R- and T-state structures of EcPfk and BsPfk, characteristic quaternary and tertiary structural changes were described for the conformational switch (2– 6). As with many other allosteric enzymes, the T to R transition is accompanied by a rotation of subunits, here by 7° around the p axis. A comparison to the PpPfk ␣ dimer shows that the subunit orientations in the yeast enzyme more closely resemble the T-state structure (Supplemental Figs. 7– 8). The structure of the region between strands I at the subunit interface across the r axis of the bacterial enzymes differ for the 2 allosteric states: in the T state, these strands are involved in direct hydrogen bonding interactions, forming a continuous  sheet between the subunits (Supplemental Fig. 9). In the R state, the strands shift apart, and a layer of water molecules mediates the interactions. In PpPfk, this interface along the r axis occurs twice between the ␣ and  chains: at the interface in the N-terminal halves, direct hydrogen bonding contacts and the distance between strands I clearly confirm the T state of the present structure (Supplemental Fig. 9). In the C-terminal halves, the situation is less clear due to the presence of P862 in strand I⬘ of the ␣ subunit and of P836 in strand I⬘ of the  chain. These proline residues break the regular strand structures and disturb the hydrogen bonding pattern. However, the close distance of the neighboring strands at the interface is also more characteristic of the T-state structure. The active site is of course also affected by the allosteric switch, mostly the F-6-P binding site, which is formed by residues of both chains. Closing of the F-6-P site inhibits the enzyme in the T state (6). In the PpPfk structure, residues of strand I of the  chain are positioned relative to the F-6-P binding site of the ␣ subunit in a way that resembles the situation in the bacterial T-state structures rather than the R state (Supplemental Fig. 10). Taken together, these findings show that PpPfk is in an inhibited T-state structure and that the general mechanism of the allosteric switch within the catalytic halves of the ␣ dimer might be similar to the bacterial enzymes. The only structural feature that contradicts the notion that the present PpPfk structure is in the T state is the azimuthal rotation between the tetramers in the upper and lower half of the enzyme. Electron microscopy studies (20, 21) on ScPfk showed that the rotational angle between the top and bottom tetramers (i.e., the angle between the r axis of the tetramer in the top half and the r axis of the tetramer in the bottom
The FASEB Journal 䡠 www.fasebj.org
STRA¨TER ET AL.
half of Fig. 1 or Fig. 2) decreases from 75° to 46° for the F-6-P-activated R state compared with the ATP-bound T state. For the homooctameric S. pombe Pfk, an azimuthal angle of 65° was determined for the F-6-Pbound state and of 50° for the ATP-bound state (44). This angle amounts to 81.5° in the present PpPfk crystal structure. In the EM structure of PpPfk analyzed in the presence of 1 mM ATP and 3 mM MgSO4, an azimuthal angle of 50° was observed. The superpositions of the N-terminal halves (Supplemental Fig. 7B) and of the C-terminal halves (Supplemental Fig. 11) of the ␣ and  chains show that the 2 chains superimpose closely when only 1 half corresponding to a bacterial subunit is matched. However, the relative orientation of the 2 halves of each chain differs by 7.7°. The corresponding rotation axis does not exactly correspond to any of the r, p, q axes, its direction is closest to the r axis. It remains to be shown whether this asymmetry between the 2 chains changes with the allosteric switch. N-terminal domains of the ␣ subunits Electron density maps of the N-terminal domains of the ␣ subunits are much weaker than those of the remaining parts of the structure and in the refined structure the B values are significantly higher (Table 1 and Supplemental Fig. 12). Since these domains are nevertheless clearly folded, they obviously undergo a rigid body motion relative to the rest of the protein. Interestingly, the pseudo-2-fold symmetry axis r, which relates the core Pfk regions (N- and C-terminal halves) of the ␣ and  chains, is not valid for the N-terminal domains. As the N-terminal domain of the ␣ chain is
located at the distal end of the molecule along the z axis, the corresponding domain of the  chain would be positioned at the octamerization interface if the r dyad is applied to this domain. Instead, it is also located at the distal end of the molecule along the z axis. The N-terminal domains of the ␣ and  chains form a tightly interacting dimer in which both  sheets of each chain assemble to 2 continuous larger sheets in the heterodimer (Fig. 5). However, the 2 N-terminal domains that form this closely interacting heterodimer are not from the same ␣ heterodimer as the rest of the chain. Instead, the ␣ chain of one ␣-heterodimer interacts with the  chain of the second ␣-heterodimer across the tetramerization interface. Via this domain swapping between 2 ␣ heterodimers, tight interactions are formed within and between the heterodimers, thus stabilizing the association of the ␣22 heterotetramer. The N-terminal domains are linked to the rest of the chain via flexible linkers of ⬃25 residues. These residues are not defined in the electron density maps and also have no predicted secondary structure. The 2 N-terminal domains of the ␣ and  chains have a similar fold (RMSD of 2.2 Å for 107 of 141/174 residues, 11% sequence identity; Supplemental Fig. 13A). Each chain forms 2 mixed 4-stranded  sheets with 2 or 3 ␣ helices on 1 side of the sheet (near strands 2 and 6). At the other edge of the sheet, the subunits dimerize such that 2 continuous sheets of 8 strands are formed. The 2  sheets form a sandwich at the dimerization interface but have a strongly curved shape such that they are far apart at the terminal strands. The ␣ helices fill most of the space between the sheets at the ends. An alignment to known protein folds in the PDB
Figure 5. Structure of the glyoxalase I-like domain formed by the N-terminal domains of the ␣ dimer. A) View along the pseudo-2-fold axis relating the ␣ and  chains in the formation of the glyoxalase dimer. The ␣ chain is shown with red helices and blue strands;  chain is shown with orange helices and light blue strands. Surface of the rest of the protein is colored as indicated. At bottom, z dyad and parts of the symmetry-related second glyoxalase-like domain (green) are visible. B) Superposition with the fosfomycin resistance protein, FosX, from Listeria monocytogenes as the most similar known protein structure, depicted as a transparent cartoon (DALI; Z⫽9.1, RMSD 3.1 Å for 109 of 131 residues, PDB 2p7p; ref. 46). Also shown are S-(N-hydroxy-N-iodophenylcarbamoyl) glutathione as a substrate analog and the catalytic zinc ion from the structure 1qin (human glyoxalase I; ref. 48). C) Domain swapping of the ␣ chains across the tetramerization interface, with view along the z dyad. N- and C-terminal halves of the catalytic Pfk core structure of the ␣-heterodimers are shown as molecular surfaces; glyoxalase-like domains are shown as cartoon ribbons. Surface and ribbon folds are colored according to the chain, as indicated by labels on protein surface (A and C, ␣ chains; B and D,  chains). X-RAY STRUCTURE OF P. PASTORIS PHOSPHOFRUCTOKINASE
95
Figure 6. Fold of the ␥ subunit. A) A ␥ subunit (cartoon with red helices and blue strands) linking ␣ and  chains across tetramerization (chain B with chain C) and octamerization interfaces (chains B and C with E). The ␣ chains are depicted as a light-brown surface and  chains as a gray surface. B) Stereo view of ␥ subunit in superposition with human catechol O-methyltransferase (PDB 3bwm, RMSD 3.1 Å for 181 of 215 residues, 15% sequence identity; ref. 52), shown as a transparent cartoon fold. Also shown are SAM (green) and 3,5-dinitrocatechol (orange) marking the active site of the superimposed enzyme.
(DALI server; ref. 45) shows that the N-terminal domains are related to glyoxalase I (E.C. 4.4.1.5, Fig. 5B; ref. 46). This enzyme catalyzes the isomerization of the spontaneously formed hemithioacetal of an ␣-oxoaldehyde and GSH to S-2-hydroxyacylglutathione derivatives. The active sites of the homodimeric enzyme consisting of a divalent metal ion and a binding pocket for the glutathione compounds are located at the interface between both domains, and they are almost completely surrounded by the curved  sheets. A superposition with substrate analogs suggests that the active site is not conserved in PpPfk (Fig. 5B). Residues 151 to 154 of the ␣ chain and 49 to 55 of the  chain fill most of the space of the former glutathione binding site. Also, the 3 metal coordinating residues are not conserved and the electron density maps do not indicate the presence of a metal ion in this region. In agreement with this finding, no glyoxalase I activity was detected for PpPfk (assayed as described in ref. 47). In addition to forming a strong interaction across the tetramerization interface by formation of the domainswapped glyoxalase dimer, the dimer also bridges the  subunits by interactions between the glyoxalase-like domain and the core Pfk regions (Fig. 5A and Supplemental Fig. 12B). From the ␣ chain A, a loop between strands 3 and 4 (residues 64 to 106) interacts with  chain B. This long loop is characterized by a complete lack of secondary structure elements. It is either not present in the glyoxalase I enzymes (e.g., PDB 2rk0; unpublished observations) or it is shorter and has a different conformation (PDB 1qin; ref. 48). Further interactions are formed between strand 2 and the following loop of chain A and with the B chain. Chain D of this glyoxalase dimer interacts with residues after strand 8 of the B chain, and the short loop between strand 6 and helix B interacts with the D chain. There are no direct interactions between the glyoxalase dimer of chains A and D with the second dimer formed by chains C and B. The closest distance is observed across 96
Vol. 25
January 2011
the z axis between the loops after strand 2, which may interact via water molecules. The presence of the glyoxalase I-related domains at the N terminus of both chains of yeast Pfks suggests that yeasts acquired this domain by a gene fusion event from an ancient glyoxalase I before the second gene duplication event that led to the ␣ and  chains (Supplemental Fig. 4). The enzymatic glyoxalase I function is not preserved, as shown by the PpPfk structure and the lack of catalytic activity. After the first gene duplication and fusion event, the ␣2 dimers could evolve into stable ␣4-tetramers, since the C-terminal halves formed a tetramerization interface. Fusion to the glyoxalase gene might have supported or initiated tetramerization by formation of the domainswapped dimers, such that the ␣ chain of one dimer forms a “glyoxalase dimer” with the ␣ chain of the other dimer across the tetramerization interface. It is possible that these domains no longer have important structural or kinetic functions in current fungal Pfks but may have instead acquired additional regulatory functions in the interaction with other proteins in yeast. A large number of phosphorylation sites found in both N-terminal domains of ScPfk supports such a function (49, 50). Proteolytic cleavage of the flexible linkers between the glyoxalase domains and the core Pfk halves in ScPfk generates 17S octamers that exhibit a similar allosteric behavior as the intact Pfk (14). The unique ␥ subunit The ␥ subunit of PpPfk consists of a single domain with an ␣/-structure (Fig. 6). A central mixed 8-stranded  sheet is flanked by 2 layers of ␣ helices on each side. With the exception of the 4 N-terminal residues and loop 152 to 175, which points toward the solvent and is presumably flexible, all residues have been modeled into the electron density maps. To analyze the evolutionary origin of the fold, we compared the structure of the ␥ subunit to the
The FASEB Journal 䡠 www.fasebj.org
STRA¨TER ET AL.
structures in the PDB using the DALI server (45). A significant match was found to S-adenosylmethionine (SAM)-dependent methyltransferases (E.C. 2.1.1.6), with catechol-O-methyltransferase (PDB 2CL5; ref. 51) as the closest homologue (16% sequence identity for 178 out of 215 residues). A superposition with cocrystal structures with bound SAM and substrates shows that neither the cosubstrate nor the substrate binding sites are conserved (Fig. 6). The comparison also shows that the largest differences between the folds of catechol-O-methyltransferase and the ␥ subunit are observed in those regions of the ␥ subunit that interact with the ␣-and  chains: the loop of residues 296 to 330, which includes helix L, interacts with the N- and C-terminal halves of the  subunit and with helix 8 of the C-terminal half of the ␣ chain. The loop of residues 200 to 223 including helix H interacts with the N- and C-terminal halves of the ␣ chain. Mainly via these 2 loops the ␥ subunit bridges 2 ␣ dimers at the tetramerization interface (Fig. 6A). Also, the octamerization interface is affected by the ␥ subunit: residues of the loops 72– 82 and 112–117 interact with region 895–930 of the ␣ chain across the octamerization interface. This loop of the ␣ chain is unique to the eukaryotic Pfks, but it is shorter and not well conserved in other Pfks or in the  chain (Supplemental Fig. 13). It is not present in the bacterial enzymes. A ␥-deletion mutant strain forms ␣44 octamers, which are enzymatically active but differ in their allosteric behavior from the wild-type enzyme (27). The mutant protein is also less stable. Apparently, Pichia acquired the unique ␥ subunit from an ancient SAM-dependent methyltransferase. The evolutionary development of the loops described above allowed for a tight binding at the interface between the N-terminal and C-terminal halves of the ␣ and  chains and between both chains, such that the ␥ chain is thus perfectly positioned to modulate the conformational switch between the R and T state, which in analogy with the bacterial Pfks is likely to include reorientations of the halves of both chains relative to each other within and between chains. EM data on ScPfk demonstrate further that the allosteric transition is accompanied by a relative rotation of the upper and lower halves of the enzyme around the z axis by 29° in the yeast enzymes (21), which will also be influenced by the ␥ subunit bridging the 2 halves across the octamerization interface. In summary, the structure of the intact full-length (␣␥)4-PpPfk provides detailed insight into the architecture of eukaryotic Pfks and the development of these complex regulated enzymes from an ancestor Pfk resembling the current bacterial enzymes. This includes the effector binding sites, the oligomer structure, and the evolutionary origin of the Nterminal domains of yeast Pfks as well as of the unique ␥ subunit in Pichia species. Our structural model clearly shows that eukaryotic Pfks and their substrate and effector sites did not simply evolve from bacterial ancestors by domain and gene amplification. An unexpected, novel ATP binding site has been characterized, whereas other binding sites such as the nucleotide binding sites in the C-terminal halves are destroyed. The structure of the T-state forms a molecular basis to further study the allosteric X-RAY STRUCTURE OF P. PASTORIS PHOSPHOFRUCTOKINASE
mechanism of eukaryotic Pfks by biochemical and structural studies, in particular concerning the Rstate structure, the conformational transition and the role of the unexpected novel ATP binding sites. The structure and evolutionary origin of the N-terminal domains raise interesting questions about their function in the current yeast Pfks as well as during the evolution of these enzymes. The Deutsche Forschungsgemeinschaft is acknowledged for funding (SFB 610). The authors thank the Joint Berlin MXLaboratory at BESSY II, the European Synchrotron Radiation Facility (Grenoble, France), and the European Molecular Biology Laboratory Hamburg Outstation at Deutsches ElektronenSynchrotron (Hamburg, Germany) for beam time and assistance during synchrotron data collection, as well as the Helmholtz Zentrum Berlin for traveling support. The authors also thank Anja Hutschenreuther for performing measurements of glyoxalase I activity of PpPfk.
REFERENCES 1. 2. 3. 4. 5. 6. 7.
8.
9.
10.
11. 12. 13.
14.
Blangy, D., Buc, H., and Monod, J. (1968) Kinetics of the allosteric interactions of phosphofructokinase from Escherichia coli. J. Mol. Biol. 31, 13–35 Evans, P. R., and Hudson, P. J. (1979) Structure and control of phosphofructokinase from Bacillus stearothermophilus. Nature 279, 500 –504 Evans, P. R., Farrants, G. W., and Hudson, P. J. (1981) Phosphofructokinase: structure and control. Philos. Trans. R Soc. Lond. B Biol. Sci. 293, 53– 62 Shirakihara, Y., and Evans, P. R. (1988) Crystal structure of the complex of phosphofructokinase from Escherichia coli with its reaction products. J. Mol. Biol. 204, 973–994 Rypniewski, W. R., and Evans, P. R. (1989) Crystal structure of unliganded phosphofructokinase from Escherichia coli. J. Mol. Biol. 207, 805– 821 Schirmer, T., and Evans, P. R. (1990) Structural basis of the allosteric behaviour of phosphofructokinase. Nature 343, 140 – 145 Riley-Lovingshimer, M. R., Ronning, D. R., Sacchettini, J. C., and Reinhart, G. D. (2002) Reversible ligand-induced dissociation of a tryptophan-shift mutant of phosphofructokinase from Bacillus stearothermophilus. Biochemistry 41, 12967–12974 Paricharttanakul, N. M., Ye, S., Menefee, A. L., Javid-Majd, F., Sacchettini, J. C., and Reinhart, G. D. (2005) Kinetic and structural characterization of phosphofructokinase from Lactobacillus bulgaricus. Biochemistry 44, 15280 –15286 Martinez-Oyanedel, J., McNae, I. W., Nowicki, M. W., Keillor, J. W., Michels, P. A., Fothergill-Gilmore, L. A., and Walkinshaw, M. D. (2007) The first crystal structure of phosphofructokinase from a eukaryote: Trypanosoma brucei. J. Mol. Biol. 366, 1185– 1198 McNae, I. W., Martinez-Oyanedel, J., Keillor, J. W., Michels, P. A., Fothergill-Gilmore, L. A., and Walkinshaw, M. D. (2009) The crystal structure of ATP-bound phosphofructokinase from Trypanosoma brucei reveals conformational transitions different from those of other phosphofructokinases. J. Mol. Biol. 385, 1519 –1533 Sols, A. (1981) Multimodulation of enzyme activity. Curr. Top. Cell Regul. 19, 77–101 Poorman, R. A., Randolph, A., Kemp, R. G., and Heinrikson, R. L. (1984) Evolution of phosphofructokinase– gene duplication and creation of new effector sites. Nature 309, 467– 469 Heinisch, J., Ritzel, R. G., von Borstel, R. C., Aguilera, A., Rodicio, R., and Zimmermann, F. K. (1989) The phosphofructokinase genes of yeast evolved from two duplication events. Gene 78, 309 –321 Kopperschla¨ger, G., Ba¨r, J., and Stellwagen, E. (1993) Limited proteolysis of yeast phosphofructokinase. Sequence locations of cleavage sites created by the actions of different proteinases. Eur. J. Biochem. 217, 527–533
97
15.
16. 17. 18.
19.
20.
21.
22.
23.
24.
25. 26.
27.
28.
29. 30. 31.
32. 33.
34.
98
Tijane, M. N., Seydoux, F. J., Hill, M., Roucous, C., and Laurent, M. (1979) Octameric structure of yeast phosphofructokinase as determined by crosslinking with disuccinimidyl beta-hydromuconate. FEBS Lett. 105, 249 –253 Telford, J. N., Lad, P. M., and Hammes, G. G. (1975) Electron microscope study of native and crosslinked rabbit muscle phosphofructokinase. Proc. Natl. Acad. Sci. U. S. A. 72, 3054 –3056 Parmeggiani, A., Luft, J. H., Love, D. S., and Krebs, E. G. (1966) Crystallization and properties of rabbit skeletal muscle phosphofructokinase. J. Biol. Chem. 241, 4625– 4637 Obmolova, G., Kopperschla¨ger, G., Heinisch, J., and Rypniewski, W. R. (1998) Crystallization and preliminary X-ray analysis of the 12S form of phosphofructokinase from Saccharomyces cerevisiae. Acta Crystallogr. D Biol. Crystallogr. 54, 96 –98 Ruiz, T., Kopperschla¨ger, G., and Radermacher, M. (2001) The first three-dimensional structure of phosphofructokinase from Saccharomyces cerevisiae determined by electron microscopy of single particles. J. Struct. Biol. 136, 167–180 Ruiz, T., Mechin, I., Ba¨r, J., Rypniewski, W., Kopperschla¨ger, G., and Radermacher, M. (2003) The 10.8-A structure of Saccharomyces cerevisiae phosphofructokinase determined by cryoelectron microscopy: localization of the putative fructose 6-phosphate binding sites. J. Struct. Biol. 143, 124 –134 Bárcena, M., Radermacher, M., Ba¨r, J., Kopperschla¨ger, G., and Ruiz, T. (2007) The structure of the ATP-bound state of S. cerevisiae phosphofructokinase determined by cryo-electron microscopy. J. Struct. Biol. 159, 135–143 Nissler, K., Hofmann, E., Stel’maschchuk, V., Orlova, E., and Kiselev, N. (1985) An electron microscopy study of the quarternary structure of yeast phosphofructokinase. Biomed. Biochim. Acta 44, 251–259 Kricke, J., Mayer, F., Kopperschla¨ger, G., and Kriegel, T. (1999) Phosphofructokinase-1 from Saccharomyces cerevisiae: analysis of molecular structure and function by electron microscopy and self-catalysed affinity labelling. Int. J. Biol. Macromol. 24, 27–35 Plietz, P., Damaschun, G., Kopperschla¨ger, G., and Mu¨ller, J. J. (1978) Small-angle x-ray scattering studies on the quaternary structure of phosphofructokinase from baker’s yeast. FEBS Lett. 91, 230 –232 Edelmann, A., and Ba¨r, J. (2002) Molecular genetics of 6-phosphofructokinase in Pichia pastoris. Yeast 19, 949 –956 Kirchberger, J., Ba¨r, J., Schellenberger, W., Dihazi, H., and Kopperschla¨ger, G. (2002) 6-phosphofructokinase from Pichia pastoris: purification, kinetic and molecular characterization of the enzyme. Yeast 19, 933–947 Tanneberger, K., Kirchberger, J., Ba¨r, J., Schellenberger, W., Rothemund, S., Kamprad, M., Otto, H., Scho¨neberg, T., and Edelmann, A. (2007) A novel form of 6-phosphofructokinase. Identification and functional relevance of a third type of subunit in Pichia pastoris. J. Biol. Chem. 282, 23687–23697 Benjamin, S., Radermacher, M., Kirchberger, J., Scho¨neberg, T., Edelmann, A., and Ruiz, T. (2009) 3D structure of phosphofructokinase from Pichia pastoris: Localization of the novel gamma-subunits. J. Struct. Biol. 168, 345–351 Cowtan, K. D., and Zhang, K. Y. (1999) Density modification for macromolecular phase improvement. Prog. Biophys. Mol. Biol. 72, 245–270 Winn, M. D., Murshudov, G. N., and Papiz, M. Z. (2003) Macromolecular TLS refinement in REFMAC at moderate resolutions. Methods Enzymol. 374, 300 –321 Blanc, E., Roversi, P., Vonrhein, C., Flensburg, C., Lea, S. M., and Bricogne, G. (2004) Refinement of severely incomplete structures with maximum likelihood in BUSTER-TNT. Acta Crystallogr. D Biol. Crystallogr. 60, 2210 –2221 Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26, 283–291 Davis, I. W., Leaver-Fay, A., Chen, V. B., Block, J. N., Kapral, G. J., Wang, X., Murray, L. W., Arendall, W. B., III, Snoeyink, J., Richardson, J. S., and Richardson, D. C. (2007) MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 35, W375–W383 Ferreras, C., Hernandez, E. D., Martinez-Costa, O. H., and Aragon, J. J. (2009) Subunit interactions and composition of the fructose 6-phosphate catalytic site and the fructose 2,6-bisphos-
Vol. 25
January 2011
35.
36.
37.
38. 39. 40. 41. 42. 43.
44. 45. 46.
47.
48.
49.
50.
51.
52.
phate allosteric site of mammalian phosphofructokinase. J. Biol. Chem. 284, 9124 –9131 Martinez-Costa, O. H., Sanchez-Martinez, C., Sanchez, V., and Aragon, J. J. (2007) Chimeric phosphofructokinases involving exchange of the N- and C-terminal halves of mammalian isozymes: implications for ligand binding sites. FEBS Lett. 581, 3033–3038 Heinisch, J. J., Boles, E., and Timpel, C. (1996) A yeast phosphofructokinase insensitive to the allosteric activator fructose 2,6-bisphosphate. Glycolysis/metabolic regulation/allosteric control. J. Biol. Chem. 271, 15928 –15933 Rais, B., Ortega, F., Puigjaner, J., Comin, B., Orosz, F., Ovadi, J., and Cascante, M. (2000) Quantitative characterization of homoand heteroassociations of muscle phosphofructokinase with aldolase. Biochim. Biophys. Acta 1479, 303–314 Roberts, S. J., and Somero, G. N. (1987) Binding of phosphofructokinase to filamentous actin. Biochemistry 26, 3437–3442 Kemp, R. G., and Gunasekera, D. (2002) Evolution of the allosteric ligand sites of mammalian phosphofructo-1-kinase. Biochemistry 41, 9426 –9430 Arvanitidis, A., and Heinisch, J. J. (1994) Studies on the function of yeast phosphofructokinase subunits by in vitro mutagenesis. J. Biol. Chem. 269, 8911– 8918 Li, Y., Rivera, D., Ru, W., Gunasekera, D., and Kemp, R. G. (1999) Identification of allosteric sites in rabbit phosphofructo1-kinase. Biochemistry 38, 16407–16412 Freyer, R. (1977) Phosphofructokinase aus Hefe, kinetisches Verhalten verschiedener Enzymformen und Diskussion eines mathematischen Modell. Habilitation thesis, University of Leipzig, Germany Rodicio, R., Strauss, A. and, Heinisch, J. J. (2000) Single point mutations in either gene encoding the subunits of the heterooctameric yeast phosphofructokinase abolish allosteric inhibition by ATP. J. Biol. Chem. 275, 40952– 40960 Benjamin, S., Radermacher, M., Ba¨r, J., Edelmann, A., and Ruiz, T. (2007) Structures of S. pombe phosphofructokinase in the F6P-bound and ATP-bound states. J. Struct. Biol. 159, 498 –506 Holm, L., Kaariainen, S., Rosenstrom, P., and Schenkel, A. (2008) Searching protein structure databases with DaliLite v. 3. Bioinformatics 24, 2780 –2781 Fillgrove, K. L., Pakhomova, S., Schaab, M. R., Newcomer, M. E., and Armstrong, R. N. (2007) Structure and mechanism of the genomically encoded fosfomycin resistance protein, FosX, from Listeria monocytogenes. Biochemistry 46, 8110 – 8120 Hollenbach, M., Hintersdorf, A., Huse, K., Sack, U., Bigl, M., Groth, M., Santel, T., Buchold, M., Lindner, I., Otto, A., Sicker, D., Schellenberger, W., Almendinger, J., Pustowoit, B., Birkemeyer, C., Platzer, M., Oerlecke, I., Hemdan, N., and Birkenmeier, G. (2008) Ethyl pyruvate and ethyl lactate down-regulate the production of pro-inflammatory cytokines and modulate expression of immune receptors. Biochem. Pharmacol. 76, 631– 644 Cameron, A. D., Ridderstrom, M., Olin, B., Kavarana, M. J., Creighton, D. J., and Mannervik, B. (1999) Reaction mechanism of glyoxalase I explored by an X-ray crystallographic analysis of the human enzyme in complex with a transition state analogue. Biochemistry 38, 13480 –13490 Albuquerque, C. P., Smolka, M. B., Payne, S. H., Bafna, V., Eng, J., Zhou, H. (2008) A multidimensional chromatography technology for in-depth phosphoproteome analysis. Mol. Cell. Proteomics 7, 1389 –1396 Smolka, M. B., Albuquerque, C. P., Chen, S. H., and Zhou, H. (2007) Proteome-wide identification of in vivo targets of DNA damage checkpoint kinases. Proc. Natl. Acad. Sci. U. S. A. 104, 10364 –10369 Palma, P. N., Rodrigues, M. L., Archer, M., Bonifacio, M. J., Loureiro, A. I., Learmonth, D. A., Carrondo, M. A., and Soares-da-Silva, P. (2006) Comparative study of ortho- and meta-nitrated inhibitors of catechol-O-methyltransferase: interactions with the active site and regioselectivity of O-methylation. Mol. Pharmacol. 70, 143–153 Rutherford, K., Le, T., I, Stenkamp, R. E., and Parson, W. W. (2008) Crystal structures of human 108V and 108M catechol O-methyltransferase. J. Mol. Biol. 380, 120 –130
The FASEB Journal 䡠 www.fasebj.org
Received for publication May 6, 2010. Accepted for publication August 26, 2010.
STRA¨TER ET AL.
