Portal de Periódicos da CAPES

Structural Basis for the Feedback Regulation of Escherichia coli Pantothenate Kinase by Coenzyme A

2000; Elsevier BV; Volume: 275; Issue: 36 Linguagem: Inglês

10.1074/jbc.m003190200

1083-351X

Mikyung Yun, Cheon‐Gil Park, Ji-Yeon Kim, Charles O. Rock, Suzanne Jackowski, Hee‐Won Park,

RNA regulation and disease

Pantothenate kinase (PanK) is a key regulatory enzyme in the coenzyme A (CoA) biosynthetic pathway and catalyzes the phosphorylation of pantothenic acid to form phosphopantothenate. CoA is a feedback inhibitor of PanK activity by competitive binding to the ATP site. The structures of the Escherichia coli enzyme, in complex with a nonhydrolyzable analogue of ATP, 5′-adenylimido-diphosphate (AMPPNP), or with CoA, were determined at 2.6 and 2.5 Å, respectively. Both structures show that two dimers occupy an asymmetric unit; each subunit has a α/β mononucleotide-binding fold with an extensive antiparallel coiled coil formed by two long helices along the dimerization interface. The two ligands, AMPPNP and CoA, associate with PanK in very different ways, but their phosphate binding sites overlap, explaining the kinetic competition between CoA and ATP. Residues Asp127, His177, and Arg243 are proposed to be involved in catalysis, based on modeling of the pentacoordinate transition state. The more potent inhibition by CoA, compared with the CoA thioesters, is explained by a tight interaction of the CoA thiol group with the side chains of aromatic residues, which is predicted to discriminate against the CoA thioesters. The PanK structure provides the framework for a more detailed understanding of the mechanism of catalysis and feedback regulation of PanK. Pantothenate kinase (PanK) is a key regulatory enzyme in the coenzyme A (CoA) biosynthetic pathway and catalyzes the phosphorylation of pantothenic acid to form phosphopantothenate. CoA is a feedback inhibitor of PanK activity by competitive binding to the ATP site. The structures of the Escherichia coli enzyme, in complex with a nonhydrolyzable analogue of ATP, 5′-adenylimido-diphosphate (AMPPNP), or with CoA, were determined at 2.6 and 2.5 Å, respectively. Both structures show that two dimers occupy an asymmetric unit; each subunit has a α/β mononucleotide-binding fold with an extensive antiparallel coiled coil formed by two long helices along the dimerization interface. The two ligands, AMPPNP and CoA, associate with PanK in very different ways, but their phosphate binding sites overlap, explaining the kinetic competition between CoA and ATP. Residues Asp127, His177, and Arg243 are proposed to be involved in catalysis, based on modeling of the pentacoordinate transition state. The more potent inhibition by CoA, compared with the CoA thioesters, is explained by a tight interaction of the CoA thiol group with the side chains of aromatic residues, which is predicted to discriminate against the CoA thioesters. The PanK structure provides the framework for a more detailed understanding of the mechanism of catalysis and feedback regulation of PanK. pantothenate kinase E. coli PanK 5′-adenylimido-diphosphate, lithium salt root mean square deviation adenosine 5′-O-(thiotriphosphate) CoA is the predominant acyl group carrier in biology and participates in a wide variety of biochemical reactions including pyruvate dehydrogenase, citrate synthase, α-ketoglutarate dehydrogenase, and the synthesis and β-oxidation of fatty acids. The universal CoA biosynthetic pathway consists of five enzymatic steps (1Abiko Y. J. Biochem. ( Tokyo ). 1967; 61: 290-299Crossref PubMed Scopus (48) Google Scholar,2Jackowski S. Neidhardt F.C. Curtiss R. Gross C.A. Ingraham J.L. Lin E.C.C. Low K.B. Mgasanik B. Reznikoff W. Riley M. Schaechter M. Umbarger H.E. Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society of Microbiology, Washington, D.C.1996: 687-694Google Scholar). The first step produces 4′-phosphopantothenic acid by phosphorylation of pantothenic acid. The subsequent formation of 4′-phosphopantetheine is a two-step process in which 4′-phosphopantothenate and cysteine are converted to 4′-phosphopantothenoyl-cysteine by formation of a peptide linkage followed by decarboxylation of the cysteine. In the final two steps, 4′-phosphopantetheine is adenylated to form dephospho-CoA, which in turn is phosphorylated at the 3′ position of ribose to form CoA. The first step of CoA biosynthesis catalyzed by pantothenate kinase (PanK)1 is a key regulatory point in CoA biosynthesis (3Jackowski S. Rock C.O. J. Bacteriol. 1981; 148: 926-932Crossref PubMed Google Scholar, 4Vallari D.S. Jackowski S. Rock C.O. J. Biol. Chem. 1987; 262: 2468-2471Abstract Full Text PDF PubMed Google Scholar, 5Karasawa T. Yoshida K. Furukawa K. Hosoki K. J. Biochem. ( Tokyo ). 1972; 71: 1065-1067Crossref PubMed Scopus (16) Google Scholar, 6Robishaw J.D. Neely J.R. Am. J. Physiol. 1984; 246: H532-H541Crossref PubMed Google Scholar, 7Fisher M.N. Neely J.R. FEBS Lett. 1985; 190: 293-296Crossref PubMed Scopus (12) Google Scholar, 8Fisher M.N. Robishaw J.D. Neely J.R. J. Biol. Chem. 1985; 260: 15745-15751Abstract Full Text PDF PubMed Google Scholar, 9Halvorsen O. Skrede S. Eur. J. Biochem. 1982; 124: 211-215Crossref PubMed Scopus (41) Google Scholar). Escherichia coli PanK (bPanK) is a homodimer of 36-kDa subunits, and the amino acid sequence contains an A-type ATP-binding consensus sequence, GXXXXGKS (10Song W.J. Jackowski S. J. Bacteriol. 1992; 174: 6411-6417Crossref PubMed Google Scholar, 11Song W.J. Jackowski S. J. Biol. Chem. 1994; 269: 27051-27058Abstract Full Text PDF PubMed Google Scholar). bPanK exhibits highly positive cooperative ATP binding and mediates a sequential ordered mechanism with ATP as the leading substrate (11Song W.J. Jackowski S. J. Biol. Chem. 1994; 269: 27051-27058Abstract Full Text PDF PubMed Google Scholar). PanK activity is inhibited by nonesterified CoA and to a lesser extent by its thioesters, which competitively interfere with ATP binding (4Vallari D.S. Jackowski S. Rock C.O. J. Biol. Chem. 1987; 262: 2468-2471Abstract Full Text PDF PubMed Google Scholar, 12Calder R.B. Williams R.S. Ramaswamy G. Rock C.O. Campbell E. Unkles S.E. Kinghorn J.R. Jackowski S. J. Biol. Chem. 1999; 274: 2014-2020Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar). A detailed mechanism of PanK inhibition by CoA is not known at the molecular level. Interestingly, nonesterified CoA is the most potent inhibitor of theE. coli enzyme (4Vallari D.S. Jackowski S. Rock C.O. J. Biol. Chem. 1987; 262: 2468-2471Abstract Full Text PDF PubMed Google Scholar), whereas acetyl-CoA is a more effective inhibitor of eukaryotic enzymes (8Fisher M.N. Robishaw J.D. Neely J.R. J. Biol. Chem. 1985; 260: 15745-15751Abstract Full Text PDF PubMed Google Scholar, 9Halvorsen O. Skrede S. Eur. J. Biochem. 1982; 124: 211-215Crossref PubMed Scopus (41) Google Scholar, 12Calder R.B. Williams R.S. Ramaswamy G. Rock C.O. Campbell E. Unkles S.E. Kinghorn J.R. Jackowski S. J. Biol. Chem. 1999; 274: 2014-2020Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar). Eukaryotic PanK cDNAs have recently been cloned, one fromAspergillus (12Calder R.B. Williams R.S. Ramaswamy G. Rock C.O. Campbell E. Unkles S.E. Kinghorn J.R. Jackowski S. J. Biol. Chem. 1999; 274: 2014-2020Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar) and two from mouse (13Rock C.O. Calder R.B. Karim M.A. Jackowski S. J. Biol. Chem. 2000; 275: 1377-1383Abstract Full Text Full Text PDF PubMed Scopus (164) Google Scholar). ADrosophila PanK has been identified as encoded by thefumble gene and may play a role in cell division (GenBankTM accession number AF221546). Comparison of the protein sequences predicted from these cDNAs with the bPanK amino acid sequence points out strong dissimilarity between prokaryotic and eukaryotic PanKs, although significant homology is found among the PanK sequences within each class (12Calder R.B. Williams R.S. Ramaswamy G. Rock C.O. Campbell E. Unkles S.E. Kinghorn J.R. Jackowski S. J. Biol. Chem. 1999; 274: 2014-2020Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar). This observation suggests that targeting the bacterial enzyme would be an effective therapeutic strategy for development of new anti-infective agents. We investigated the molecular basis for bPanK catalysis and regulation by determining the structures of binary complexes of bPanK and AMPPNP, a nonhydrolyzable analogue of ATP, and CoA. The CoaA gene encoding bPanK (10Song W.J. Jackowski S. J. Bacteriol. 1992; 174: 6411-6417Crossref PubMed Google Scholar) was subcloned into pET21a (Novagen, Inc.) and transformed intoE. coli B834(DE3)pLysS (Novagen, Inc.). The E. coli cells harboring the pET21-bPanK plasmid were grown in 4 liters of M9 minimal medium (6 g/liter Na2HPO4, 3 g/liter KH2 PO4, 0.5 g/liter NaCl, and 1 g/liter NH4 Cl) containing 0.4% glucose, 1 mm MgSO4, 0.1 mmCaCl2, 0.0005% thiamine, amino acid mixture (0.5 g/liter of each amino acid), 50 μg/ml of ampicillin, 34 μg/ml of chloramphenicol, and 0.1 g/liter of methionine at 37 °C. When the culture reached an absorbance of 0.7–0.8 at 600 nm, the cells were harvested and washed with the M9 minimal medium twice. The cells were resuspended in the M9 minimal medium containing 40 mg/l of seleno-l -methionine (Sigma). After induction with 0.4 mmisopropyl-1-thio-β-d-galactopyranoside, the cells were allowed to grow for another 3 h to be harvested. The harvested cell pellet was resuspended in 400 ml of disruption buffer (20 mm Tris-HCl, pH 8.0, 1 mm dithiothreitol, 1 mm EDTA, and 1 mm phenylmethylsufonyl fluoride) and lysed with sonication. The cell lysate was centrifuged at 14,000 × g for 30 min. The supernatant was purified by Q-Sepharose anion exchange chromatography in which bPanK was eluted with an increasing NaCl gradient. Eluted factions containing bPanK were pooled and concentrated. The concentrated sample was then purified with a S-200 Superdex gel filtration column for the fast protein liquid chromatography system (Amersham Pharmacia Biotech). The elution buffer was 20 mm Tris-HCl, pH 7.6, 200 mm NaCl, 1 mm dithiothreitol, and 1 mm EDTA, and the flow rate was 60 ml/h. Fractions containing bPanK were pooled and dialyzed at 4 °C for 15 h against dialysis buffer (20 mmTris-HCl, pH 8.0, 1 mm dithiothreitol, and 1 mmEDTA). The dialyzed sample was concentrated to 20 mg/ml for crystallization trials. Co-crystals of selenomethionine bPanK and AMPPNP were grown by the hanging drop vapor diffusion method. Before crystallization, AMPPNP was added to the protein to a final concentration of 2 mm, and the mixture was incubated overnight at 4 °C. The hanging drops containing the equal volume of the protein and the reservoir solution were equilibrated at 4 °C against the reservoir solution (11% polyethylene glycol 8,000 and 0.1m HEPES, pH 7.5). The trigonal crystals appeared after several days, which grew further for about 1 more week. These crystals were transferred to a cryoprotectant solution containing 40% glycerol, 15% polyethylene glycol 8,000, 2 mm dithiothreitol, 0.5 mm AMPPNP, 2.5 mm MgCl2, and 0.1m HEPES at pH 7.5 for cryo-data collection. Crystals of bPanK in complex with CoA were grown at room temperature by the hanging drop vapor diffusion method. To achieve the full occupancy of bound CoA, bPanK was mixed with CoA at the final concentration of 2 mm, and the mixture was incubated for 24 h at 4 °C. Equal volumes of protein solution and reservoir solution were combined. The reservoir solution consisted of 10% polyethylene glycol 4,000, 50 mm Li2SO4, 2% isopropanol, and 0.1m N-[2-acetamido]-2-iminodiacetic acid at pH 6.5. Rod-like crystals appeared after 1 week and grew for another 2–3 weeks. These crystals were frozen using a cryoprotectant solution containing 40% glycerol, 12% polyethylene glycol 4,000, 2.5 mm dithiothreitol, 0.1 mm CoA, 2 mmMgCl2, 50 mm Li2SO4, and 0.1 m N-[2-acetamido]-2-iminodiacetic acid at pH 6.5 for cryo-data collection. Multiwavelength anomalous dispersion data were measured from a single crystal of the AMPPNP-bound bPanK at beamline X12C at the National Synchrotron Light Source, equipped with a Q4 CCD detector (Brandeis) operating at 100 K. Three wavelengths near the selenium absorption edge, 0.95 Å (the remote), 0.9791 Å (the peak), and 0.9794 Å (the inflection point), were chosen by measuring the x-ray absorption spectrum of the protein crystal. The data set from each wavelength was integrated, reduced, and scaled by the programs DENZO and SCAPEPACK (14Otwinowski Z. Minor W. Methods Enzymol. 1997; 276A: 307-326Crossref Scopus (38782) Google Scholar). The crystals of the AMPPNP-bound enzyme belong to space group P3221 with unit cell dimensions ofa = 130.1 Å and c = 281.8 Å. The program SOLVE (15Terwilliger T. Berendzen J. Acta Crystallogr. Sect. D Biol. Crystallogr. 1999; 55: 501-505Crossref PubMed Scopus (49) Google Scholar) was used to scale three wavelength data sets together, to locate selenium positions and to calculate the multiwavelength anomalous dispersion phases. The statistics of data collection and phasing are summarized in TableI. The initial electron density map calculated at 3.0 Å resolution was improved by the noncrystallographic symmetry averaging method using the program DM (the CCP4 suite) (16Collaborative Computational Project Number 4 Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Crossref PubMed Scopus (19879) Google Scholar). The atomic model was built using the graphics programs O (17Jones T.A. Zou J.Y. Cowan S.W. Kjeldgaard M. Acta Crystallogr. Sect. A. 1991; 47: 110-119Crossref PubMed Scopus (13055) Google Scholar) and XtalView (18McRee D.E. XtalView (3.2.1). The Scripps Research Institute, La Jolla, CA1998Google Scholar), followed by crystallographic refinement in the program XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar). The refinement steps included rigid body, simulated annealing, conjugate gradient minimization, and individual B-factor refinement. A solvent mask was calculated with the bulk solvent correction routine in XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar). During refinement, NCS restraints were imposed on the main chain atoms. The final structure contains four subunits of bPanK (6–316 residues in one subunit and 8–316 residues in the other three subunits). The refinement statistics are shown in TableII.Table IData collection and phasing statisticsData sourceAMPPNP-bound enzymeCoA-bound enzymeInflectionPeakRemoteWavelength (Å)0.97950.97910.950.95Resolution (Å)50.0–2.650.0–2.650.0–2.850.0–2.5Number of reflections Measured966,9611,100,377800,291335,425 Unique83,23784,98565,77546,593Completeness (%, > −3ς) Overall97.499.595.891.2 Last shell79.695.871.552.7R sym(%)1-aR sym = Σhkl[Σi‖Ihkl,i − ‖]/Σhkl,i , where Ihkl,i is the intensity of an individual measurement of the reflection with Miller indices h, h, and l and is the mean intensity of that reflection.8.67.75.83.9Average I/ς(I)13.315.117.315.6Anomalous scattering factors f′ values−7.37−6.47−3.11 f values3.784.164.01Phasing statistics Mean figure of merit0.5 RMS (FH)/RMS (E)1-bRMS (FH)/RMS (E) is the ratio of the root mean square value of the calculated heavy atom structure factor (FH) to the root mean square value of the difference between calculated and observed derivative structure factors (E), where it is averaged not only over all reflections but over all phases for each reflection, weighted by the phase probability. Centric reflections1.2 Acentric reflections1.11-a R sym = Σhkl[Σi‖Ihkl,i − ‖]/Σhkl,i , where Ihkl,i is the intensity of an individual measurement of the reflection with Miller indices h, h, and l and is the mean intensity of that reflection.1-b RMS (FH)/RMS (E) is the ratio of the root mean square value of the calculated heavy atom structure factor (FH) to the root mean square value of the difference between calculated and observed derivative structure factors (E), where it is averaged not only over all reflections but over all phases for each reflection, weighted by the phase probability. Open table in a new tab Table IIRefinement statisticsAMPPNP-bound enzymeCoA-bound enzymeRefinement Resolution (Å)50.0–2.650.0–2.5 R free2-aR work = Σ∥F obs‖ − ‖F calc∥/Σ‖F obs‖, where ‖F obs‖ and ‖F calc‖ are observed and calculated structure factor amplitudes, respectively.0.2600.258 R work2-bR free is equivalent to R work except that 5% of the total reflections were set aside for an unbiased test of the progress of refinement.0.2160.204 Number of reflections (F > 2 ς (F))72,35341,211 Number of protein atoms10,0649,980 Number of ligand atoms128192 Number of water molecules306301Stereochemistry Rmsd bond length (Å)0.0090.008 Rmsd bond angles (degrees)1.661.59 Average B factors (Å2) Protein main-chain/side-chain42.5/44.331.3/32.7 Ligand39.932.8Residues from Ramachandran plot Most favored regions (%)85.888.9 Additional allowed regions (%)12.09.8 Generously allowed regions (%)2.31.32-a R work = Σ∥F obs‖ − ‖F calc∥/Σ‖F obs‖, where ‖F obs‖ and ‖F calc‖ are observed and calculated structure factor amplitudes, respectively.2-b R free is equivalent to R work except that 5% of the total reflections were set aside for an unbiased test of the progress of refinement. Open table in a new tab Complete diffraction data from a single crystal of the CoA-bound bPanK were also measured at beamline X12C at the National Synchrotron Light Source. The programs DENZO and SCALEPACK (14Otwinowski Z. Minor W. Methods Enzymol. 1997; 276A: 307-326Crossref Scopus (38782) Google Scholar) were used to integrate and scale the data set. The crystals of the CoA-bound enzyme belong to space group P1 with unit cell dimensions ofa = 62.0 Å, b = 71.2 Å,c = 87.7 Å, α = 102.4°, β = 89.5°, and γ = 93.2°. The data collection statistics are summarized in Table I. The structure of the CoA-bound bPanK was determined by the molecular replacement method using one subunit of the AMPPNP-bound enzyme as a search model. The cross-rotation and translation functions and Patterson correlation refinement were calculated using the program XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar). Using the data between 15.0 and 4.0 Å, the rotation function, followed by Patterson correlation refinement, gave four outstanding solutions that correspond to four subunits of bPanK. Because any point can be taken as an origin in space group P1, rotation of the initial search model according to one of the four rotation function solutions determined the orientation and position of the first subunit. With fixing this orientation and position of the first subunit, the orientations of the other three subunits were determined by applying the corresponding noncrystallographic symmetry operations to the first subunit. The positions of the other three subunits were then determined by finding the relative x, y, andz translations of each of the three subunits with respect to the first subunit. The model including all four subunits was subjected to rigid body refinement at a resolution between 6. 0 and 4. 0 Å in the program XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar). The high resolution limit, which restricted the maximum shift of structure in the refinement, was set to 4. 0 Å resolution to provide a possible large shift of the structure up to 4.0 Å. The low resolution limit was set to 6.0 Å resolution below which the intensities of reflections were severely affected by the diffraction of solvent molecules in the crystal. The Rfactor at this stage was high (51.0%), suggesting that the conformation of the CoA-bound structure was different from that of the search model, the AMPPNP-bound structure. To consider the solvent contribution to the intensities of reflections at low resolution, the structure factor amplitudes of solvent molecules in the crystal were calculated using the bulk solvent correction routine in XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar), and thus the low resolution limit could be extended to 50.0 Å resolution. Crystallographic refinement including simulated annealing, conjugate gradient minimization, and individual B-factor refinement was then performed at the resolution between 50.0 and 2.5 Å. After the first round of refinement, the quality of the structure was dramatically improved with a working R factor of 27.4% and a freeR factor of 30.1%. The F o −F c difference map showed continuous density for CoA, verifying the correct molecular replacement solution. Several iterations of model building in the program O (17Jones T.A. Zou J.Y. Cowan S.W. Kjeldgaard M. Acta Crystallogr. Sect. A. 1991; 47: 110-119Crossref PubMed Scopus (13055) Google Scholar) and the refinement in the program XPLOR (19Brünger A.T. X-PLOR , , version 3.1. Yale University Press, New Haven, CT1992Google Scholar) further dropped both R factors. The final refinement statistics are shown in Table II. Because of the slight difference between four subunits, noncrystallographic symmetry restraints were applied to only 89% of total residues of each subunit. In the final model, residues 1–5 in subunit 1; residues 1–7 and 211–212 in subunit 2; residues 1–5 and 210–213 in subunit 3; and residues 1–7, 83–85, and 212–214 in subunit 4 are excluded because they are disordered in the crystal. In the structures of the AMPPNP-bound and the CoA-bound bPanKs, four identical subunits are found in an asymmetric unit. When the four subunits occupying an asymmetric unit of each structure are compared, the average RMSD of Cα atoms for the AMPPNP-bound and the CoA-bound enzymes are 0.46 and 0.34 Å, respectively. These values are consistent with estimated errors in their atomic coordinates. A subunit of bPanK adopts a mononucleotide-binding fold (20Schulz G.E. Curr. Biol. 1992; 2: 61-67Crossref Scopus (308) Google Scholar): a seven-stranded β-sheet (strands, 2, 3, and 8–11) is flanked by α-helices (D and E on one side and G and J on the other side) (Fig.1). An intervening loop between strand 2 and helix E, known as the P-loop, contains most of the residues that interact with phosphate oxygens of the AMPPNP. There are four small antiparallel β-strands (strands 4–7) that are not part of the main β-sheet (Fig. 1 a). Residues from helices H and I including their intervening loop are involved in binding CoA (Fig.1 a). The N-terminal region including strand 1, helices A and D, and a loop between helices A and B forms the major part of the dimer interface (Fig. 1 a). The Dali server (21Holm L. Sander C. Nucleic Acids Res. 1997; 25: 231-234Crossref PubMed Scopus (362) Google Scholar) was used to search for known proteins structurally similar to bPanK. Applied to one subunit of bPanK, the best matches occurred with proteins containing a mononucleotide binding fold,i.e. phosphoribulokinase from Rhodobacter sphaeroides (Protein Data Bank code 1a7j, RMSD 1.9 Å, 185 residues), 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase from rat (Protein Data Bank code 1bif, RMSD 1.7 Å, 96 residues), and adenylate kinase from Bacillus stearothermophilus (Protein Data Bank code 1zin, RMSD 2.1 Å, 110 residues). In light of the wide distribution of the mononucleotide binding fold, these findings are not surprising. However, the remarkable similarity of bPanK to R. sphaeroides phosphoribulokinase is unexpected because their sequence homology is limited to the P loop. Phosphoribulokinase catalyzes the ATP-dependent phosphorylation of ribulose 5-phosphate to ribulose 1,5-bisphosphate in the Calvin cycle (22Wolosiuk R.A. Ballicora M.A. Hagelin K. FASEB J. 1993; 7: 622-637Crossref PubMed Scopus (61) Google Scholar). Four subunits found in an asymmetric unit of the AMPPNP-bound enzyme are assembled to form two dimers, consistent with the fact that bPanK functions as a dimer in solution (11Song W.J. Jackowski S. J. Biol. Chem. 1994; 269: 27051-27058Abstract Full Text PDF PubMed Google Scholar). The dimerization interface occurs between two long α-helices from each of the subunits that form an extensive antiparallel coiled coil (Fig.2 a). The interface contact including this coiled coil buries 4118 Å2 of accessible surface area of each subunit. The physical makeup of the interface contact is slightly hydrophobic: 57% of nonpolar and 43% of polar/charged residues. By comparison, the dimer-dimer interface is not extensive. About 1790 Å2 of the accessible surface area of each dimer is buried. Most of the residues involved in this contact are predominantly hydrophilic: 27% nonpolar and 73% polar/charged residues. There are also two dimers in the asymmetric unit of the CoA-bound bPanK. However, the dimerization interface of the CoA-bound enzyme is not identical to that of the AMPPNP-bound enzyme resulting in a conformational change. The dimerization interface, depicted parallel to helix D of both subunits, shows a helical break where a couple of turns of helix D of one subunit, involving residues 68–73, are unwound (Fig. 2 b). This helical break is observed in both dimers. Consequently, the last two turns of helix D of the first subunit are shifted closer to helix D of the second subunit. This shift disturbs the dimerization interface to bring about an 18° rotation of the second subunit with respect to the first subunit, resulting in movement of the second subunit of as much as 15 Å (Fig.2 c). The altered interface contact buries 3004 Å2 of accessible surface area of each subunit. The two dimers also interact by burying of 2174 Å2 of each dimer surface. The electron density of AMPPNP is clearly defined in both dimers occupying an asymmetric unit of the AMPPNP-bound bPanK (Fig. 3 a). AMPPNP is bound in a groove formed by residues from the P-loop, the connecting loop of helices B and C, and the connecting loop of strands 10 and 11. AMPPNP with residues lining the binding site are shown in Fig.3 b. The adenine base is sandwiched by the side chains of Asn43 and His307, the latter side chain further interacting with the side chain of Trp239. The N6 nitrogen of the adenine base interacts via water-mediated hydrogen bonds with the hydroxyl groups of Tyr55 and Thr104 and the amino group of Lys303, whereas its N1 nitrogen forms a water-mediated hydrogen bond to the amide nitrogen of Ser47. The ribose is hydrogen-bonded to the carboxyl group of Asp45. The α-phosphate interacts with the hydroxyl group and the amide nitrogen of Thr103. The β- and γ-phosphates interact with the amino group of Lys101 and the β-phosphate interacts with the amide nitrogens of Gly100, Lys101, and Ser102, whereas the γ-phosphate interacts with the amide nitrogen of Ala98 and the guanidinium group of Arg243. A magnesium ion is coordinated by β- and γ-phosphates on one side and the carboxyl group of Glu199 and the hydroxyl group of Ser102 on the other side. The well defined electron density for CoA is seen in both dimers occupying the asymmetric unit of the CoA-bound bPanK (Fig. 4 a). CoA is bound in the enzyme in a bent conformation in a deep pocket lined by residues from helix H, the P loop, the connecting loop of strands 5 and 6 and the connecting loop of helices H and I (Fig. 4 b). Because CoA is a competitive inhibitor with respect to ATP and these two ligands share the ADP moiety, it was suggested that both ligands bind to the same site (4Vallari D.S. Jackowski S. Rock C.O. J. Biol. Chem. 1987; 262: 2468-2471Abstract Full Text PDF PubMed Google Scholar, 11Song W.J. Jackowski S. J. Biol. Chem. 1994; 269: 27051-27058Abstract Full Text PDF PubMed Google Scholar). Unexpectedly, the adenine base of CoA is inserted between the side chains of His177 and Phe247, whereas the adenine-interacting residues in the AMPPNP-bound enzyme are Asn43 and His307 (Fig.4 b). The α-phosphate of CoA is salt bridged to the guanidinium group of Arg243 and the amino group of Lys101 that also interacts with the β-phosphate of CoA (Fig. 4 b). The 3′-phosphate of CoA is hydrogen-bonded to the amide nitrogen of Ile42 and the hydroxyl group of Ser102 and salt bridged to the guanidinium group of Arg106 (Fig. 4 b). These interactions of the 3′-phosphate group may be essential for the inhibitory effect of CoA because dephospho-CoA is a significantly less potent inhibitor of bPanK activity (4Vallari D.S. Jackowski S. Rock C.O. J. Biol. Chem. 1987; 262: 2468-2471Abstract Full Text PDF PubMed Google Scholar). By comparison, the same 3′-phosphate group in several CoA-binding proteins shows no interaction with the protein and is exposed to solvent (for review see Ref. 23Engel C. Wierenga R. Curr. Opin. Struct. Biol. 1996; 6: 790-797Crossref PubMed Scopus (71) Google Scholar). The exceptions are acyl-CoA binding protein (24Kragelund B.B. Andersen K.V. Madsen J.C. Knudsen J. Poulsen F.M. J. Mol. Biol. 1993; 230: 1260-1277Crossref PubMed Scopus (126) Google Scholar) and the surfactin synthetase activating enzyme Sfp (25Reuter K. Mofid M.R. Marahiel M.A. Ficner R. EMBO J. 1999; 18: 6823-6831Crossref PubMed Scopus (179) Google Scholar), in which the 3′-phosphate interacts with the protein residues such as His and Lys. Hydrophobic atoms of the pantothenate moiety of CoA form van der Waals' contacts with residues Leu130, Tyr175, and Ile281 (Fig. 4 b). The carbonyl oxygen of the pantothenate moiety near the β-mercaptoethylamine moiety is hydrogen bonded to the hydroxyl group of Tyr240 and to the amide group of Asn282 (Fig. 4 b). The amide nitrogen of the β-mercaptoethylamine moiety is hydrogen-bonded to the hydroxyl group of Tyr180 (Fig. 4 b). Strikingly, the thiol group of the β-mercaptoethylamine moiety is tightly sealed from water molecules by interacting with four aromatic residues, Phe244, Phe252, Phe259, and Tyr262 and also by intra-molecular hydrogen bonding to the amino group of the adenine base (Fig. 4 b). The thiol group approaches the face of Phe259 at the distance