Crystal Structure of the Phosphoenolpyruvate-binding Enzyme I-Domain from the Thermoanaerobacter tengcongensis PEP: Sugar Phosphotransferase System (PTS)

doi:10.1016/j.jmb.2004.11.077

J. Mol. Biol. (2005) 346, 521–532

Crystal Structure of the Phosphoenolpyruvate-binding Enzyme I-Domain from the Thermoanaerobacter tengcongensis PEP: Sugar Phosphotransferase System (PTS) Anselm Erich Oberholzer†, Mario Bumann†, Philipp Schneider Christoph Ba¨chler, Christian Siebold, Ulrich Baumann* and Bernhard Erni* Department of Chemistry and Biochemistry, University of Berne, Freiestrasse 3, CH-3012 Bern, Switzerland

Enzyme I (EI), the first component of the phosphoenolpyruvate (PEP): sugar phosphotransferase system (PTS), consists of an N-terminal proteinbinding domain (EIN) and a C-terminal PEP-binding domain (EIC). EI transfers phosphate from PEP by double displacement via a histidine residue on EIN to the general phosphoryl carrier protein HPr. Here, we ˚ crystal structure of the homodimeric EIC domain from report the 1.82 A Thermoanaerobacter tengcongensis, a saccharolytic eubacterium that grows optimally at 75 8C. EIC folds into a (ba)8 barrel with three large helical insertions between b2/a2, b3/a3 and b6/a6. The large amphipathic ˚ 2 of accessible surface area per monomer. dimer interface buries 3750 A A comparison with pyruvate phosphate dikinase (PPDK) reveals that the active-site residues in the empty PEP-binding site of EIC and in the liganded PEP-binding site of PPDK have almost identical conformations, pointing to a rigid structure of the active site. In silico models of EIC in complex with the Z and E-isomers of chloro-PEP provide a rational explanation for their difference as substrates and inhibitors of EI. The EIC domain exhibits 54% amino acid sequence identity with Escherichia coli and 60% with Bacillus subtilis EIC, has the same amino acid composition but contains additional salt-bridges and a more complex salt-bridge network than the homology model of E. coli EIC. The easy crystallization of EIC suggests that T. tengcongensis can serve as source for stable homologs of mesophilic proteins that are too labile for crystallization. q 2004 Elsevier Ltd. All rights reserved.

*Corresponding authors

Keywords: phosphotransferase system; phosphoenolpyruvate; thermophilic; bacteria; PEP-utilising enzyme

Introduction The bacterial phosphoenolpyruvate:sugar phosphotransferase system (PTS) plays a dual role in

transport and phosphorylation of sugars and hexitols on one hand and regulation of the cellular metabolism in response to the availability of these carbohydrates on the other. The molecular basis of

† A.E.O. and M.B. contributed equally to this work. Present addresses: C. Ba¨chler, ZLB Behring AG, 3000 Bern, Switzerland; C. Siebold, Division of Structural Biology, Wellcome Trust Centre for Human Genetics, University of Oxford, Rosevelt Drive, Oxford OX3 7BN, UK. Abbreviations used: PEP, phosphoenolpyruvate; Z- and E-ClPEP, (Z)-3- and (E)-3-chlorophosphoenolpyruvate; P-pyr, 3-phosphonopyruvate (O3P–CH2–CO–COOH); PTS, PEP:sugar phosphotransferase system; Prv, pyruvate; EI, enzyme I of the PTS; EIN, N-terminal domain of EI; EIC, C-terminal domain of EI; PPDK, pyruvate phosphate dikinase; SeMet, selenomethionine; NTA, nitrilo triacetic acid; MAD, multi-wavelength anomalous dispersion; C, Cys; D, Asp; E, Glu; N, Asn; R, Arg. E-mail addresses of the corresponding authors: [email protected]; [email protected] 0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

522 its multiple functions is a protein phosphorylation cascade comprising four phosphoprotein units that transfer phosphoryl groups sequentially from phosphoenolpyruvate to the transported carbohydrates.1 Enzyme I (EI) is the first protein at the top of the divergent cascade. It transfers phosphate from PEP to HPr, the general high-energy phosphate carrier protein of the PTS.2,3 HPr itself then serves as phosphoryl donor to the various PTS transporters (EII, enzymes II) of different and sometimes overlapping sugar specificity. These transporters consist of three protein subunits or domains. The EIIA and EIIB domains serve as a phosphate transfer relay to the carbohydrate substrate that is being translocated by the membranespanning EIIC domain.4 The phosphorylation state of some EIIA and EIIB subunits also serves as “input” for the PTS-mediated signal transduction. Steady-state phosphorylation of EIIA and EIIB is high in the absence and low during transport of the cognate sugars. Phosphoryl transfer occurs through phosphohistidine and phosphocysteine intermediates. PTS of highly variable size occur in a large proportion of eubacteria5 but not in eucaryotes and archaebacteria. EI of the PTS, a 64 kDa two-domain protein, is one of the best-conserved bacterial proteins with only minimal similarity to animal proteins.6,7 The amino-terminal domain (EIN) contains a histidine residue which by double displacement transfers phosphate from PEP to HPr. The three-dimensional crystal structure of EIN from Escherichia coli has been elucidated, and its mode of interaction with HPr characterized by NMR spectroscopy.8,9 EIN is composed of the HPr-binding a-helical subdomain and an a/b subdomain that bears the phosphorylatable histidine. The carboxy-terminal domain (EIC) contains the PEP-binding site.10,11 The a/b subdomain of EIN and EIC exhibit 30% sequence identity with the respective domains of the pyruvate phosphate dikinase (PPDK, SwissProt accession no. P22983, PDB code 1KC7).12 In view of its importance for the bacterial physiology, EI, mostly of E. coli, has been the object of numerous microbiological, biochemical, biophysical studies. However, the mode of action of E. coli EI and how its activity is controlled is not fully understood. EI is a dimer and EIC is responsible for dimer formation. The concentration of EI in E. coli is of the order of 5 mM.13 Dimerization of E. coli EI in vitro is strongly temperaturedependent,14 the dissociation constant shifting from 20 mM at 6 8C to 0.9 mM at 30 8C. The monomer–dimer equilibrium depends on divalent cations (Mg2C or Mn2C) as well as PEP, and the association rate constant is two to three orders of magnitude slower than usual for dimeric proteins, suggesting that dimerization in vitro is accompanied by major conformational rearrangements of the interacting EIC domains.15–18 Studies with pyrene-labeled EI point to a correlation between activity and dimerization.16,17 The activity of E. coli EI diluted to a rate-limiting concentration

Structure of EIC of the PTS from T. tengcongensis

is enhanced several-fold by the addition of an EI mutant that, by itself, is catalytically inactive.19,20 Whereas the structure of the N-terminal domain of E. coli EI has been determined,9 the structures of EIC or full-length protein are not yet known. E. coli EIC is proteolytically unstable, and the unfolding transition temperature of the C-terminal domain in full-length EI varies between 41 8C in the absence of substrates and 60 8C in the H189A mutant in the presence of PEP and Mg2C.21,22 Attempts to express E. coli EIC met with mixed success. 23,24 The marginal stability of E. coli EIC and the predictability of its structure based on its sequence similarity with the PEP-binding (ba)8 barrel domain of pyruvate phosphate dikinase12,25–27 may have further discouraged crystallization. Here, we present the X-ray structure of the EIC domain from Thermoanaerobacter tengcongensis, a Gramnegative, saccharolytic bacterium that grows optimally at 75 8C,28,29 and we discuss the implications of this structure for the understanding of EI activity and its inhibition.

Results Cloning, purification and crystallization of EIC from Thermoanaerobacter tengcongensis The known dynamic behavior of E. coli EI may compromise the formation of well-ordered threedimensional crystals. To find a more stable EI variant we subcloned the genes from Staphylococcus aureus, Bacillus stearothermophilus, Borrelia burgdorferi, E. coli, Enterococcus faecalis and Haemophilus influenzae and overexpressed the proteins in E. coli. All EIs complemented PTS activity in an E. coli DptsI mutant and, after purification, displayed in vitro phosphotransferase activity, indicating that the proteins were active. EIntr of E. coli, a paralogue of EI containing, in addition, a putative sensor domain,30 was expressed in inclusion bodies only. The recombinant EI-like domain of the multiphosphoryl protein FruB of Pseudomonas aeruginosa could not be expressed in E. coli. At first EI of S. aureus appeared to be the most suitable of all. To prevent the protein from aggregation and to improve homogeneity, all surface-exposed cysteine residues were exchanged for alanine and serine. One out of 12 mutant proteins afforded crystals that ˚ but crystal diffracted to a resolution of 3.5 A mosaicity and disorder of the crystal lattice prevented interpretation (results not shown31). T. tengcongensis is the first, and so far the only, thermophilic eubacterium possessing a PTS and, since thermophilic orthologues have frequently proven their use in structural biology in the past, we focused our efforts on EI from this organism. When EI with a C-terminal histidine tag was expressed in E. coli, only fragments and almost no full-length protein were obtained. Four major fragments that were retained by Ni2C-NTA affinity chromatography all started with a methionine

523

Structure of EIC of the PTS from T. tengcongensis

residue, which hinted at internal translation initiation after rare codons.32 The most abundant fragment (O50%) turned out to be the complete EIC domain. Thereafter codons 252 to 573 were PCRamplified and the recombinant EIC domain was expressed in an E. coli Rosetta strain that supplements tRNAs for rare codons. EIC could be purified to homogeneity by Ni2C-NTA and gelfiltration chromatography. It is a stable homodimer with an unfolding transition temperature of 90 8C (V. Navdaeva & B. E., unpublished results). While wild-type EIC yielded crystals of mediocre quality, SeMet-labeled EIC containing 16 SeMet molecules (0.45 SeMet/kDa) crystallized in an orthorhombic space-group suitable for MAD experiments and structure refinement. Crystal structure determination Crystals of EIC belonged to the orthorhombic ˚ , bZ91.8 A ˚ , cZ space-group P212121 (aZ82.8 A ˚ ) with two physiological dimers in the 185.9 A asymmetric unit. Only reflections with lZ4n of the (00l) reflections were strong, and a native Patterson ˚ resolution showed strong peaks (20% map at 4.5 A of the origin peak) on the sections wZ0.25 and wZ 0.5. The crystals diffracted to a maximum resolution ˚ (Table 1). The structure of EIC was solved of 1.82 A by multi wavelength anomalous dispersion (MAD)

˚ resolution. Using the peak wavelength at 2.8 A alone resulted in better correlation coefficients in the wrong space-group P21212 than in the true spacegroup, but no interpretable electron density was obtained. Using the correct space-group and FA values yielded eventually a solution with a correlation coefficient of 70% and an interpretable electron density. ˚ data Refinement of the model against 1.82 A resulted in reasonable R-factors and satisfying model geometry (Table 1). A total of 1199 residues are in the most-favored and in the additionaly allowed regions of the Ramachandran plot. The four residues (0.3%), in the disallowed region, namely Ala359 of the four monomers, are well defined in the electron density. Residues 251–572 of the EIC monomer are well ordered in all monomers, only the N-terminal Met250 in one of the four monomers is disordered. Overall structure EIC features a (ba)8 barrel fold (Figure 1). The four crystallographically independent monomers of the asymmetric unit are very similar. The RMS deviation of the Ca positions over the entire length ˚ . Only two solvent-exposed of a molecule is 0.35 A loops (residues 302–308 and 337–351) have more ˚ and 1.5 A ˚, than average RMS deviations of 0.7 A

Table 1. Data collection and refinement statistics

Space group Unit cell parameters ˚) a (A ˚) b (A ˚) c (A Molecules/a.s.u.a Data collection (XDS) Beamline ˚) Wavelength (A ˚) Resolution rangeb (A No. observations No. unique reflections Completeness (%) Rsymd (%) I/sI FOMe MAD (overall) FOMf solvent flattened Refinement (REFMAC) ˚) Resolution range (A Reflections working set Reflections test set Non-hydrogen atoms Solvent water molecules R/Rfree (%) ˚) RMSD bond lengths (A RMSD bond angles (deg.) a b c d

Low remote

Peak

P212121

P212121

82.10 91.43 181.86 4 BM14 0.9878 40–1.82 (1.94–1.82) 820,735 123,055 99.9 (99.7) 6.7 (24.4) 16.5 (4.6)

Inflection

Remote

82.82 91.85 185.96 4 BM14 0.9788 40–2.8 (2.97–2.80) 275,471c 66,731c 98.8 (92.9) 6.1 (18.4) 13.8 (5.1) 0.57 0.77

BM14 0.9791 40–2.8 (2.97–2.80) 277,294c 67,336c 99.7 (98.4) 6.3 (20.3) 13.5 (4.8)

BM14 0.8856 40–2.8 (2.97–2.80) 275,169c 66,882c 99.0 (94.3) 7.1 (23.4) 12.2 (4.2)

30–1.82 120,674 2245 10958 837 20.5/24.4 0.015 1.870

a.u., asymmetric unit. The values in parentheses of resolution range, completeness, Rsym and I/s(I) correspond to the outermost-resolution shell. Friedel pairs were treated as different
Preflections.  P P P Rsym Z hkl j jIðhkl; jÞK hIðhklÞij= K hkl j hIðhklÞi , where I(hkl;j) is the jth measurement of the intensity of the unique reflection,

(hkl) and hI(hkl)i is the mean over all symmetry-related measurements. e Figure-of-merit as computed by SOLVE. f Figure-of-merit as computed by RESOLVE.

524

Structure of EIC of the PTS from T. tengcongensis

Figure 1. Stereoview of the T. tengcongensis EIC domain. Cartoon representation of the monomer structure. The three extensions are colored red (residues 296–309), blue (residues 333–365) and green (residues 452–479). The active-site Cys502 is shown in yellow sticks.

respectively. These loops (Figure 1, magenta) are engaged in crystal contacts in two of the four monomers. The (ba)8 core has one extra N-terminal a-helix and three extensions between b2/a2, b3/a3 and b6/a6 on the C-terminal face of the barrel (Figure 2(a)). The first extension (296–309) has an irregular structure, the second extension (333–365)

comprises two short a-helices, and the third (452–479) one extra a-helix. The three extensions ˚ above the core of the protrude approximately 30 A barrel. The active-site Cys502 (see below) is located in the b7/a7 turn and is accessible over the short b8/a8 and b1/a1 turns at the rim of the barrel. EIC has the shape of an “easy chair”, the three extensions forming the backrest and the C-terminal end of the barrel the cushion (Figure 2(a)). The contact area between the EIC subunits in the physiological dimer is formed mostly by back-toback contacts between the b3/a3 and b6/a6 extensions (the backrest, Figure 2(b)). Additional contacts are provided by the b4/a4 and b5/a5 turns and the first half of helix 6. The contact area comprises 45 hydrophobic and hydrophilic ˚ 2 of monomer surface residues, which cover 3750 A (Figure 2(c)). Thus 15% of the monomer surface becomes buried upon dimer formation. The inter˚ 2 per subunit contact area of more than 3700 A monomer is exceptionally large for a dimerization domain, which usually are much smaller and of the ˚ 2.33 order of 1200–2400 A Structure comparison with pyruvate phosphate dikinase

Figure 2. Surface representation of EIC (a) monomer and (b) dimer. The three extensions are colored in red, blue, green and orange, light blue, yellow and green, respectively. (a) The monomer is obtained by a 908 rotation of the cartoon model around the horizontal (x) axis of Figure 1. (c) Intersubunit contact area of the EIC monomer. Hydrophobic residues (Ala, Val, Cys, Leu, Ile, Phe, Tyr, Trp, Pro, Met) are colored yellow, basic residues (Arg, Lys, His) blue, acidic amino acids (Glu and Asp) red, polar residues (Asn, Gln, Ser, Thr, Gly) cyan. (d) Eight-center ion-pair network on the surface of EIC. Basic residues are colored blue and acidic residues red.

EIC exhibits sequence similarity with the PEPbinding domain of pyruvate phosphate dikinase (PPDK) and PEP synthase, another mechanistically related, PEP-binding enzyme.34 PPDK catalyzes the reversible conversion of ATP, inorganic phosphate (Pi), and pyruvate to AMP, pyrophosphate, and phosphoenolpyruvate (PEP), respectively. Structures of full-length PPDK, alone (PDB entry 1DIK) and of the complex with the substrate analogue 3phosphonopyruvate (P-pyr) (PDB entry 1KC7), have been solved.12,25,26 EIC and the C-terminal PEP/pyruvate-binding domains of PPDK display 27% sequence identity and are the closest structural homologues in the Protein Data Bank. An overlay of

Structure of EIC of the PTS from T. tengcongensis

the (b/a)8 folds of EIC and the C-terminal domain of PPDK is shown in Figure 3(a). The DALI server yields a Z-score of 31.0 and an RMS deviation of ˚ for 320 residues of EIC and 279 out of 360 2.2 A amino acid residues of PPDK. Using the program Indonesia (D. Madsen et al., unpublished results), 268 residues from EIC can be superimposed with the enzyme–ligand complex between PPDK and ˚ for the paired Ca P-pyr. The RMS deviation is 1.6 A atoms. Figure 4 shows the structure-based sequence alignment between EIC and the pyruvate/PEPbinding domain of PPDK. EIC shares with PPDK the three extensions that, however, are slightly shorter. In PPDK the b2/a2 extension features two a helices (instead of one) in a helical hairpin and the b3/a3 and b6/a6 extensions also have more a-helical structure than EIC. EIC and PPDK have in common the large intersubunit contact area. Of the 45 residues making up the dimer interfaces, 32 are conserved.

525 The ligand-free PEP-binding site of EIC and the liganded site of PPDK at the C-terminal end of the b/a barrel are almost completely superimposable (Figure 3(b)). The structural details of the PPDK binding site have been characterized with crystals of PPDK in complex with P-pyr.25 The C]O and CH2–PO3 groups of P-pyr isosterically replace the C]CH2 and the –O–PO3 groups of PEP. A rich network of interactions ensures the precise positioning of the P-pyr ligand in the PPDK-binding site.25 In EI of E. coli, the PEP-binding site has been identified by affinity labeling of Cys502 with the mechanism-based inhibitor ZClPEP.10 The superimposition of the PPDK/P-pyr structure with EIC shows that all the residues that form the charge network with P-pyr and Mg2C are also conserved in EIC (Figure 3(b)). Even the side-chain positions and the relative orientations of the active-site residues are remarkably well conserved, independently of whether a substrate is bound. P-pyr can be

Figure 3. Comparison of EIC with the PEP-binding domain of Clostridium symbiosum PPDK. (a) Stereoview of backbone alignment of EIC (red) and PPDK (blue). (b) Stereoview of the superimposed PEP-binding sites of EIC and PPDK. The EIC site (red) is empty, the PPDK site (blue) contains phosphonopyruvate (P-pyr) and Mg2C. The residues of EIC are labeled in bold face and those of PPDK are in parentheses. P-pyr is shown as a stick model. Atomic colors are as follows: oxygen, red; carbon, white; phosphorus, yellow; and magnesium, magenta. The PPDK coordinates are from PDB accession number 1KC7.25

526

Structure of EIC of the PTS from T. tengcongensis

Figure 4. Structure-based sequence alignment of EIC and PPDK. Conserved residues involved in PEP/pyruvate and Mg2C binding are colored blue. The colored bar below the sequence shows the dimer interface. A red bar symbolizes residues involved only in the EIC dimer interface, a blue bar residues of the PPDK dimer interface and a green bar residues involved in both dimer interfaces.

accommodated at the C-terminal end of the b/abarrel of the T. tengcongensis EIC structure without sterical clashes and constraints. Comparison of thermostability and salt-bridges of EIC with EIC of E. coli Thermal stability of the E. coli EIC domain varies between 408 and 60 8C, depending on the experimental conditions,21,35,36 while T. tengcongensis EIC unfolds at 90 8C (V. Navdaeva & B. E., unpublished results). Electrostatic interactions are thought to act as an important factor conferring thermostability to proteins.37,38 This opinion is supported by the increased number of salt-bridges found in many structures of thermostable proteins. To compare the number and pattern of salt-bridges between EIC of T. tengcongensis and E. coli, a homology-based structure of the E. coli EIC domain was constructed on the basis of the experimental T. tengcongensis

structure. The salt-bridges present in the two structures are listed in Table 2. EIC of T. tengcongensis exhibits ten simple salt-bridges between isolated pairs of oppositely charged residues. Two arginyl, one lysyl and five glutamyl residues form a salt-bridge network consisting of seven links between opposite charges (Figure 2(d)). The modeled structure of E. coli EIC exhibits only six simple salt-bridges and the charge network consists of only five links between two arginyl and four glutamyl groups. Although the net free energy contribution of salt-bridges is difficult to quantify,39 the increased number of ionic interactions is consistent with the increased thermostability of the T. tengcongensis EIC. Properties other than salt-bridges are very similar between the two domains (Table 2). They have essentially the same amino acid composition and the same number of charged residues. EIC of T. tengcongensis contains five extra proline residues in non-conserved

527

Structure of EIC of the PTS from T. tengcongensis

Table 2. Comparison of structural features of EIC from T. tengcongensis and E. coli T. tencongensis Salt-bridge network

Simple salt-bridges

Charged residues

Lys371 Lys371 Arg375 Arg375 Arg375 Arg379 Arg379 Arg304 Arg332 Arg358 Arg400 Lys410

Glu408 Glu412 Glu310 Glu412 Glu416 Glu310 Glu311 Glu312 Glu298 Asp339 Glu445 Glu407

Arg465 Lys471 Lys486 Arg538 Lys546 15 Arg 27 Lys

Glu504 Glu468 Asp490 Glu251 Glu543 20 Asp 34 Glu 302 5.01 3748 22 24

Hydrogen bonds PI ˚ 2) Accessible surface area buried at dimer interface (A Non-polar residues at interface Polar residues at interface

sequences, a higher Lys content, and 10% more hydrogen bonds than predicted for EIC of E. coli. A single Asn287-Gly288 sequence, a linkage thought to be weak and hydrolysis-sensitive,40 is conserved in both proteins.

Discussion Enzyme I of E. coli, the key enzyme of the PTS, has been the object of numerous physiological, biochemical and biophysical studies. By using EIC from a thermophilic bacterium, stability problems encountered with the EIC domain of E. coli could be circumvented. EIC is a physiological homodimer ˚ 2. with a contact area of 3700 A In view of this large contact interface, it is not clear how a dimer–monomer transition14,16,17 could occur during the catalytic cycle without the exposition of large patches of hydrophobic residues. Could the dimer–monomer transition observed with E. coli EI be contingent on the marginal thermal stability of its EIC domain? The contact areas in the EIC dimer of T. tencongensis and in the modeled dimeric structures of E. coli are comparable with respect to area and hydrophobicity and thus do not offer an explanation for the different stabilities. EIC displays strong structure similarity with PPDK. Particularly striking is the similarity between the PEP-binding sites. They display very similar side-chain orientations, although PPDK is complexed with a substrate analogue while EIC is without a ligand. From this, it appears that ligand binding alone should not cause the significant conformational changes that have been observed with EI of E. coli.15–18 Changes in backbone and side-chain conformation of EIC possibly occur only

E. coli (model)

Arg375 Arg375 Arg375 Arg379 Arg379 Arg304

Glu310 Glu412 Glu416 Glu310 Glu311 Glu312

Arg358

Asp339

Lys410 Arg414 Arg465

Glu407 Glu422 Glu504

Lys486

Asp490

Lys546 21 Arg 18 Lys

Glu543 20 Asp 33 Glu 276 4.94 3053 19 21

upon complexation with the EIN domain, to which the phosphoryl group is transferred. Modeling of the EIC–EIN complex A preliminary model of the complex between the two domains was obtained by computationally docking EIN of E. coli EI (PDB accession code 1ZYM) with EIC of T. tengcongensis. Predicted complexes were filtered by a distance restraint between N3 of His189 of EIN and the phosphorus atom of P-pyr modeled in the active site of EIC. This distance becomes minimal during in-line phosphate transfer from PEP to His189. With full-length EIN as a rigid docking partner, the minimal approach between phosphorus and His189 was always ˚ (data not shown). When the a/b greater than 20 A swivel sub-domain instead of EIN was used, the ˚ , which could be minimal distance was 8.1 A ˚ by small changes in the sidereduced to 7.4 A chain torsion angles of His189 (Figure 5(a) and (b)). This is still twice the P–N distance expected of the transition state comprising a pentacoordinated phosphoryl group in a trigonal bipyramidal geometry,41 and it suggests that other adjustments in backbone and side-chain conformation in the vicinity of the PEP-binding and/or the P acceptor site are necessary for docking with EIN. Modeling of the EIC–ClPEP complexes The reaction mechanism at the PEP-binding site of E. coli EI has been characterized with C-3modified PEP analogues Z-ClPEP (Z: COOH and Cl trans to C]C double bond) and E-ClPEP.10,19 Z-ClPEP functions as substrate and mechanismbased inhibitor of EI, which targets the active-site

528

Structure of EIC of the PTS from T. tengcongensis

Figure 5. Hypothetical docking models. (a) Stereoview of docking between EIC of T. tengcongensis and the a/b swivel domain of E. coli EI. Ribbon representation of EIC (cyan) with phosphonopyruvate (sticks, P-pyr) modeled in the PEPbinding site according to Figure 3. The five best-fitting a/b domains are shown in backbone representations (blue, red, white, yellow and violet), the two a and b structures surrounding the active-site His189 (in sticks) are shown in ribbon representation. (b) Close-up view of His189 and P-pyr in the docking model. Atomic colors are as follows: oxygen, red; carbon, white; phosphorus, yellow.

Cys502. E-ClPEP, in contrast, has no significant activity. The two ClPEP isomers were modeled in the EIC-binding site with the carboxylate group, C2 and phosphate group in the same position as those of P-Pyr in the PPDK model (Figure 6). Under these constraints, the two isomers differ by the orientation of the chlorine atom at C-3. E-ClPEP can be accomodated in the active site without sterical clashes (Figure 6(a)), while the chlorine atom of Z-ClPEP overlaps with the guanidino group of Arg332 (Figure 6(b)). However, the negatively polarized chlorine atom of E-ClPEP is in van der Waals contact with the carboxylate group of Glu431, while the chlorine atom of Z-ClPEP is directed towards the positively charged guanidino group of Arg332. The electrostatic repulsion of the former and attraction of the latter could thus explain the observed preference of EI for Z-ClPEP over E-ClPEP. Because of the steric interference between the chlorine atom and Arg332, C-3 of Z-ClPEP is likely to be turned away from the general acid/base Cys502. This should slow the protonation of the enolate group by the general acid. The enolate group, which is stabilized by electron dispersion by the electron-withdrawing chlorine atom, can dissociate from the binding pocket and is then

protonated by water, explaining the observed nonstereospecific protonation of the Z-isomer.19

Materials and Methods Protein expression and purification T. tengcongensis full-length EI and the N-terminal His6tagged construct of the C-terminal domain of enzyme I (EIC(251–573)) were made by cloning the PCR product derived from a T. tengcongensis genomic DNA library into the NdeI and BamHI sites of the pET28a vector (Novagen) using the primers 5 0 -ACGTACATATGGAAGGAT TAAAGCAGTTAAAAG-3 0 and 5 0 -ACGTAGGATCCT TAGCCAATATCTTTTATCACG-3 0 . EIC was expressed in Rosettae(DE3) cells (Novagen). Cells were grown at 37 8C to an A550 of 0.8, and the expression was induced by 0.2 mM isopropyl-b-D-thiogalactopyranoside (IPTG). After 12 hours induction, cells were harvested and resuspended in 20 mM Tris–HCl (pH 8.0), 400 mM NaCl, 10 mM b-mercaptoethanol, 0.2 mM PMSF and disrupted using a French press. Cell debris and the membrane fraction were removed by low-speed and high-speed centrifugation, respectively (ten minutes at 12,000g and 90 minutes at 150,000g). EIC was purified on a 20 ml Ni-NTA SUPERFLOW column (Qiagen) according to the manufacturer’s instructions. Typical yields were

529

Structure of EIC of the PTS from T. tengcongensis

Figure 6. Stereoview of the EIC active site in complex with E-ClPEP and Z-ClPEP. ClPEP and relevant side-chains are shown as stick models. The atomic colors are: oxygen, red; carbon, white; phosphorus and sulfur, yellow. 30 mg/l of pure EIC. The eluate was concentrated by Centriprep-30 (Millipore) to a volume of 2 ml. The protein was further purified by gel-filtration using a Superdex 75 (Amersham Bioscience) column equilibrated in 10 mM Hepes (pH 7.5), 400 mM NaCl, 2 mM DTT. The purification steps were monitored by SDS-PAGE. The peak fractions were pooled and dialysed against 5 mM Hepes (pH 7.5), 2 mM DTT, 5 mM CaCl2 for two hours at 4 8C and concentrated to a final concentration of 10 mg/ml. Selenomethionine (SeMet)-labeled protein was prepared by the method of methionine biosynthesis inhibition.42 Cells were grown at 37 8C to an A550 of 0.6 and the expression was induced with 0.2 mM IPTG. After 12 hours of induction at 37 8C, cells were harvested and the protein purified as described above.

100 mm!100 mm!110 mm. Only the SeMet-labeled crystals were suitable for structure determination. Before data collection, crystals were flash-cooled in a nitrogen stream at 110 K after raising the concentration of MPD of the crystallization solution to 35% (v/v). Data were collected in a MAD experiment at the Grenoble ESRF MAD beamline BM14 at 100 K, employing a Mosaic 225 CCD detector (Marresearch GmbH, Hamburg, Germany). Due to the alignment of the crystals in the cryoloop, only 0.2 8/frame could be collected. All datasets were integrated and scaled with XDS.43,44 Peak, inflection and high-remote data were collected using one crystal. For refinement purposes, a high-resolution low-remote data set was collected on a different, larger crystal. The two crystals turned out to be non-isomorphous. Data statistics are given in Table 1.

Crystallization and data collection

Structure solution, refinement and analysis

Trigonal crystals of EIC were obtained within four days by the sitting-drop, vapor-diffusion method. Drops were set up by mixing 3 ml of protein solution (10 mg/ml) with 3 ml of reservoir solution (0.1 M sodium acetate trihydrate (pH 4.6), 12% (w/v) PEG 4000) and equilibrated against 120 ml of reservoir solution at 20 8C. Crystals of SeMet-EIC were obtained within four days by mixing 2 ml of protein solution (10 mg/ml) with 2 ml of reservoir solution (0.1 M sodium acetate trihydrate (pH 4.6), 12% PEG 4000, 7% (v/v) methyl-2,4-pentanediol (MPD)). They belong to the orthorhombic space ˚ , bZ group P212121 with cell parameters of aZ82.81 A ˚ , cZ185.96 A ˚ , and contain four monomers per 91.85 A asymmetric unit. Typical crystals have an average size of

Of 64 expected selenium positions, 60 were determined employing the program SHELXD45 and FA values, i.e. the contribution of the anomalous scatterers to the structure amplitudes, computed by XPREP (G. Sheldrick, unpublished program). Phases were computed using SOLVE46 and improved by RESOLVE.47 Automatic model building was done by RESOLVE. Of 324 residues, 142 were built by the program. Further attempts to use the low-remote, high-resolution data for automated model building failed, even after transporting the electron density into the unit cell of the high-resolution crystal using MOLREP.48 Therefore, the remaining residues of one monomer were placed manually into the MAD-derived electron density map using the program O,49 version

530 9.0.3. This initial model was then placed into the unit cell of the low-remote data set by means of the program MOLREP,48 followed by density modification, phase extension and automatic model building using ArpWarp,50 version 6.0. Refinement was effected using REFMAC,51 version 5.1.24. Refinement statistics are given in Table 1. Structure-based sequence alignment was done by the program Indonesia (D. Madsen et al., unpublished results). Figure 3 was prepared using ESPript.52 Comparison of structures was effected using DALI53 and LSQMAN.54 The structures of EIC of E. coli and Mycoplasma capricolum were modeled using the SWISS MODEL server,55–57 using EIC of T. tengcongensis as template. Three-dimensional structures of E-ClPEP and Z-ClPEP were created using the Dundee PRODRG2 server,58 and were superimposed on P-pyr, an inhibitor of pyruvate phosphate dikinase25 using O. Molecular modeling was performed with the program 3D-Dock.59,60 Structure Figures were created using the program PYMOL (www.pymol.org).

Acknowledgements This research was supported by grant 3100A0105247 from the Swiss National Science Foundation and the Ciba-Geigy Jubila¨umsstiftung. A Roche Research Foundation fellowship to A.E.O. is gratefully acknowledged. We are very grateful to Dr Martin Walsh from the beamline BM14 at the European Synchrotron Radiation Facility in Grenoble.

References 1. Postma, P. W., Lengeler, J. W. & Jacobson, G. R. (1996). Phosphoenolpyruvate: carbohydrate phosphotransferase systems. In Escherichia coli and Salmonella: Cellular and Molecular Biology (Neidhardt, F. C. et al., eds), pp. 1149–1174, ASM Press, Washington, DC. 2. Chauvin, F., Brand, L. & Roseman, S. (1996). Enzyme I: the first protein and potential regulator of the bacterial phosphoenolpyruvate:glycose phosphotransferase system. Res. Microbiol. 147, 471–479. 3. Ginsburg, A. & Peterkofsky, A. (2002). Enzyme I: the gateway to the bacterial phosphoenolpyruvate: sugar phosphotransferase system. Arch. Biochem. Biophys. 397, 273–278. 4. Robillard, G. T. & Broos, J. (1999). Structure/function studies on the bacterial carbohydrate transporters, enzymes II, of the phosphoenolpyruvate-dependent phosphotransferase system. Biochim. Biophys. Acta, 1422, 73–104. 5. Reizer, J., Charbit, A., Reizer, A. & Saier, M. H. (1996). Novel phosphotransferase system genes revealed by bacterial genome analysis: operons encoding homologues of sugar-specific permease domains of the phosphotransferase system and pentose catabolic enzymes. Genome Sci. Technol. 1, 53–75. 6. LiCalsi, C., Crocenzi, T. S., Freire, E. & Roseman, S. (1991). Sugar transport by the bacterial phosphotransferase system. Structural and thermodynamic domains of enzyme I of Salmonella typhimurium. J. Biol. Chem. 266, 19519–19527.

Structure of EIC of the PTS from T. tengcongensis

7. Saier, M. H., Simoni, R. D. & Roseman, S. (1970). Physiological behaviour of enzyme I and HPr mutants of a bacterial phosphotransferase system. J. Biol. Chem. 245, 5870–5873. 8. Garrett, D. S., Seok, Y. J., Liao, D. I., Peterkofsky, A., Gronenborn, A. M. & Clore, G. M. (1997). Solution structure of the 30 kDa N-terminal domain of enzyme I of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system by multidimensional NMR. Biochemistry, 36, 2517–2530. 9. Liao, D. I., Silverton, E., Seok, Y. J., Lee, B. R., Peterkofsky, A. & Davies, D. R. (1996). The first step in sugar transport: crystal structure of the amino terminal domain of enzyme I of the E-coli PEP: sugar phosphotransferase system and a model of the phosphotransfer complex with HPr. Structure, 4, 861–872. 10. Garcia-Alles, L. F., Flu¨kiger, K., Hewel, J., Gutknecht, R., Siebold, C., Schu¨rch, S. & Erni, B. (2002). Mechanism-based inhibition of enzyme I of the Escherichia coli phosphotransferase system - cysteine 502 is an essential residue. J. Biol. Chem. 277, 6934–6942. 11. Seok, Y. J., Lee, B. R., Gazdar, C., Svenson, I., Yadla, N. & Peterkofsky, A. (1996). Importance of the region around glycine-338 for the activity of enzyme I of the Escherichia coli phosphoenolpyruvate: sugar phosphotransferase system. Biochemistry, 35, 236–242. 12. Herzberg, O., Chen, C. C., Kapadia, G., McGuire, M., Carroll, L. J., Noh, S. J. & Dunaway-Mariano, D. (1996). Swiveling-domain mechanism for enzymatic phosphotransfer between remote reaction sites. Proc. Natl Acad. Sci. USA, 93, 2652–2657. 13. Rohwer, J. M., Meadow, N. D., Roseman, S., Westerhoff, H. V. & Postma, P. W. (2000). Understanding glucose transport by the bacterial phosphoenolpyruvate:glycose phosphotransferase system on the basis of kinetic measurements in vitro. J. Biol. Chem. 275, 34909–34921. 14. Kukuruzinska, M. A., Turner, B. W., Ackers, G. K. & Roseman, S. (1984). Subunit association of enzyme I of the Salmonella typhimurium phosphoenolpyruvate: glycose phosphotransferase syste. Temperature dependence and thermodynamic properties. J. Biol. Chem. 259, 11679–11681. 15. Chauvin, F., Toptygin, D., Roseman, S. & Brand, L. (1992). Time-resolved intrinsic fluorescence of enzyme I. The monomer/dimer transition. Biophys. Chem. 44, 163–173. 16. Chauvin, F., Brand, L. & Roseman, S. (1994). Sugar transport by the bacterial phosphotransferase system. Characterization of the Escherichia coli enzyme I monomer/dimer equilibrium by fluorescence anisotropy. J. Biol. Chem. 269, 20263–20269. 17. Chauvin, F., Brand, L. & Roseman, S. (1996). Sugar transport by the bacterial phosphotransferase system. Characterization of the Escherichia coli enzyme I monomer/dimer transition kinetics by fluorescence anisotropy. J. Biol. Chem. 269, 20270–20274. 18. Han, M. K., Walbridge, D. G., Knutson, J. R., Brand, L. & Roseman, S. (1987). Nanosecond time-resolved fluorescence kinetic studies of the 5,5 0 -dithiobis(2nitrobenzoic acid) reaction with enzyme I of the phosphoenolpyruvate:glycose phosphotransferase system. Anal. Biochem. 161, 479–486. 19. Garcia-Alles, L. F., Alfonso, I. & Erni, B. (2003). Enzyme I of the phosphotransferase system: inducedfit protonation of the reaction transition state by Cys-502. Biochemistry, 42, 4744–4750.

531

Structure of EIC of the PTS from T. tengcongensis

20. Seok, Y. J., Zhu, P. P., Koo, B. M. & Peterkofsky, A. (1998). Autophosphorylation of enzyme I of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system requires dimerization. Biochem. Biophys. Res. Commun. 250, 381–384. 21. Dimitrova, M. N., Szczepanowski, R. H., Ruvinov, S. B., Peterkofsky, A. & Ginsburg, A. (2002). Interdomain interaction and substrate coupling effects on dimerization and conformational stability of enzyme I of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system. Biochemistry, 41, 906–913. 22. Dimitrova, M. N., Peterkofsky, A. & Ginsburg, A. (2003). Opposing effects of phosphoenolpyruvate and pyruvate with Mg(2C) on the conformational stability and dimerization of phosphotransferase enzyme I from Escherichia coli. Protein Sci. 12, 2047–2056. 23. Brokx, S. J., Talbot, J., Georges, F. & Waygood, E. B. (2000). Enzyme I of the Phosphoenolpyruvate:Sugar phosphotransferase system. In vitro intragenic complementation: the roles of Arg126 in phosphoryl transfer and the C-terminal domain in dimerization. Biochemistry, 39, 3624–3635. 24. Fomenkov, A., Valiakhmetov, A., Brand, L. & Roseman, S. (1998). In vivo and in vitro complementation of the N-terminal domain of enzyme I of the Escherichia coli phosphotransferase system by the cloned C-terminal domain. Proc. Natl Acad. Sci. USA, 95, 8491–8495. 25. Herzberg, O., Chen, C. C., Liu, S., Tempczyk, A., Howard, A., Wei, M. et al. (2002). Pyruvate site of pyruvate phosphate dikinase: crystal structure of the enzyme-phosphonopyruvate complex, and mutant analysis. Biochemistry, 41, 780–787. 26. Wei, M., Li, Z., Ye, D. M., Herzberg, O. & DunawayMariano, D. (2000). Identification of domain-domain docking sites within Clostridium symbiosum pyruvate phosphate dikinase by amino acid replacement. J. Biol. Chem. 275, 41156–41165. 27. Xu, Y., Yankie, L., Shen, L., Jung, Y. S., Mariano, P. S., Dunaway-Mariano, D. & Martin, B. M. (1995). Location of the catalytic site for phosphoenolpyruvate formation within the primary structure of Clostridium symbiosum pyruvate phosphate dikinase 1. Identification of an essential cysteine by chemical modification with [1-14C]bromopyruvate and sitedirected mutagenesis. Biochemistry, 34, 2181–2187. 28. Bao, Q., Tian, Y., Li, W., Xu, Z., Xuan, Z., Hu, S. et al. (2002). A complete sequence of the T. tengcongensis genome. Genome Res. 12, 689–700. 29. Xue, Y., Xu, Y., Liu, Y., Ma, Y. & Zhou, P. (2001). Thermoanaerobacter tengcongensis sp. nov., a novel anaerobic, saccharolytic, thermophilic bacterium isolated from a hot spring in Tengcong, China. Int. J. Syst. Evol. Microbiol. 51, 1335–1341. 30. Powell, B. S., Court, D. L., Inada, T., Nakamura, Y., Michotey, V., Cui, X. et al. (1995). Novel proteins of the phosphotransferase system encoded within the rpoN operon of Escherichia coli. Enzyme IIANtr affects growth on organic nitrogen and the conditional lethality of an erats mutant. J. Biol. Chem. 270, 4822–4839. 31. Siebold, C. (2002). Structural studies on proteins of the bacterial phosphotransferase system. Thesis, University of Bern. 32. Hess, D. & Hensel, R. (1996). The 3-phosphoglycerate kinase of the hyperthermophilic archaeum Pyrococcus woesei produced in Escherichia coli: loss of heat

33.

34.

35.

36.

37. 38.

39.

40. 41.

42.

43.

44. 45. 46. 47.

resistance due to internal translation initiation and its restoration by site-directed mutagenesis. Gene, 172, 121–124. Ponstingl, H., Henrick, K. & Thornton, J. M. (2000). Discriminating between homodimeric and monomeric proteins in the crystalline state. Proteins: Struct. Funct. Genet. 41, 47–57. Pocalyko, D. J., Carroll, L. J., Martin, B. M., Babbitt, P. C. & Dunaway Mariano, D. (1990). Analysis of sequence homologies in plant and bacterial pyruvate phosphate dikinase enzyme I of the bacterial phosphoenolpyruvate sugar phosphotransferase system and other pep-utilizing enzymes identification of potential catalytic and regulatory motifs. Biochemistry, 29, 10757–10765. Ginsburg, A., Szczepanowski, R. H., Ruvinov, S. B., Nosworthy, N. J., Sondej, M., Umland, T. C. & Peterkofsky, A. (2000). Conformational stability changes of the amino terminal domain of enzyme I of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system produced by substituting alanine or glutamate for the active-site histidine 189: Implications for phosphorylation effects. Protein Sci. 9, 1085–1094. Nosworthy, N. J., Peterkofsky, A., Ko¨nig, S., Seok, Y. J., Szczepanowski, R. H. & Ginsburg, A. (1998). Phosphorylation destabilizes the amino-terminal domain of enzyme I of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system. Biochemistry, 37, 6718–6726. Perutz, M. F. & Raidt, H. (1975). Stereochemical basis of heat stability in bacterial ferredoxins and in haemoglobin A2. Nature, 255, 256–259. Karshikoff, A. & Ladenstein, R. (2001). Ion pairs and the thermotolerance of proteins from hyperthermophiles: a “traffic rule” for hot roads. Trends Biochem. Sci. 26, 550–556. Bosshard, H. R., Marti, D. N. & Jelesarov, I. (2004). Protein stabilization by salt bridges: concepts, experimental approaches and clarification of some misunderstandings. J. Mol. Recogn. 17, 1–16. Vieille, C. & Zeikus, G. J. (2001). Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for thermostability. Microbiol. Mol. Biol. Rev. 65, 1–43. Wang, G. S., Louis, J. M., Sondej, M., Seok, Y. J., Peterkofsky, A. & Clore, G. M. (2000). Solution structure of the phosphoryl transfer complex between the signal transducing proteins HPr and IIAGlucose of the Escherichia coli phosphoenolpyruvate:sugar phosphotransferase system. EMBO J. 19, 5635–5649. Van Duyne, G. D., Standaert, R. F., Karplus, P. A., Schreiber, S. L. & Clardy, J. (1993). Atomic structures of the human immunophilin FKBP-12 complexes with FK506 and rapamycin. J. Mol. Biol. 229, 105–124. Kabsch, W. (1993). Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallog. 26, 795–800. Kabsch, W. (1988). Automatic indexing of rotation diffraction patterns. J. Appl. Crystallog. 21, 67–72. Schneider, T. R. & Sheldrick, G. M. (2002). Substructure solution with SHELXD. Acta Crystallog. sect. D, 58, 1772–1779. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallog. sect. D, 55, 849–861. Terwilliger, T. C. & Berendzen, J. (1999). Discrimination of solvent from protein regions in native

532

48. 49.

50.

51.

52.

53.

Structure of EIC of the PTS from T. tengcongensis

Fouriers as a means of evaluating heavy-atom solutions in the MIR and MAD methods. Acta Crystallog. sect. D, 55, 501–505. Vagin, A. & Teplyakov, A. (1997). MOLREP: an automated program for molecular replacement. J. Appl. Crystallog. 30, 1022–1025. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 101–109. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Automated protein model building combined with iterative structure refinement. Nature Struct. Biol. 6, 458–463. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallog. sect. D, 53, 240–255. Gouet, P., Robert, X. & Courcelle, E. (2003). ESPript/ ENDscript: extracting and rendering sequence and 3D information from atomic structures of proteins. Nucl. Acids Res. 31, 3320–3323. Holm, L. & Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123–138.

54. Kleywegt, G. J. & Jones, T. A. (1997). Detecting folding motifs and similarities in protein structures. Methods Enzymol. 277, 525–545. 55. Guex, N. & Peitsch, M. C. (1997). SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis, 18, 2714–2723. 56. Peitsch, M. C. (1995). Protein modeling by E-mail. Bio/Technology, 13, 658–660. 57. Schwede, T., Kopp, J., Guex, N. & Peitsch, M. C. (2003). SWISS-MODEL: an automated protein homology-modeling server. Nucl. Acids Res. 31, 3381–3385. 58. van Aalten, D. M., Bywater, R., Findlay, J. B., Hendlich, M., Hooft, R. W. & Vriend, G. (1996). PRODRG, a program for generating molecular topologies and unique molecular descriptors from coordinates of small molecules. J. Comput. Aided Mol. Des. 10, 255–262. 59. Aloy, P., Moont, G., Gabb, H. A., Querol, E., Aviles, F. X. & Sternberg, M. J. (1998). Modelling repressor proteins docking to DNA. Proteins: Struct. Funct. Genet. 33, 535–549. 60. Sternberg, M. J., Aloy, P., Gabb, H. A., Jackson, R. M., Moont, G., Querol, E. & Aviles, F. X. (1998). A computational system for modelling flexible protein– protein and protein–DNA docking. Proc. Int. Conf. Intell. Syst. Mol. Biol. 6, 183–192.

Edited by R. Huber (Received 18 October 2004; received in revised form 29 November 2004; accepted 30 November 2004)