Molecular Mechanism of Allosteric Substrate Activation in a
Thiamine Diphosphate-dependent Decarboxylase
*
□SReceived for publication, July 24, 2007, and in revised form, September 10, 2007 Published, JBC Papers in Press, September 28, 2007, DOI 10.1074/jbc.M706048200 Wim Verse´es‡§1, Stijn Spaepen¶2, Martin D. H. Wood储3, Finian J. Leeper储, Jos Vanderleyden¶, and Jan Steyaert‡§
From the‡Department of Ultrastructure, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium, the§Department of
Molecular and Cellular Interactions, VIB, Pleinlaan 2, B-1050 Brussels, Belgium, the¶Centre of Microbial and Plant Genetics, INPAC, Katholieke Universiteit Leuven, B-3001 Heverlee, Belgium, and the储University Chemical Laboratory, Lensfield Road,
Cambridge CB2 1EW, United Kingdom
Thiamine diphosphate-dependent enzymes are involved in a wide variety of metabolic pathways. The molecular mechanism behind active site communication and substrate activation, observed in some of these enzymes, has since long been an area of debate. Here, we report the crystal structures of a phenylpyru-vate decarboxylase in complex with its substrates and a covalent reaction intermediate analogue. These structures reveal the reg-ulatory site and unveil the mechanism of allosteric substrate activation. This signal transduction relies on quaternary struc-ture reorganizations, domain rotations, and a pathway of local conformational changes that are relayed from the regulatory site to the active site. The current findings thus uncover the molecular mechanism by which the binding of a substrate in the regulatory site is linked to the mounting of the catalytic machinery in the active site in this thiamine diphosphate-dependent enzyme.
Thiamine diphosphate (ThDP),4the biologically active form
of vitamin B1, is an essential cofactor for a wide variety of enzymes that mainly mediate the making and breaking of car-bon-carbon bonds adjacent to a carbonyl group (Fig. 1) (1). Although ThDP-dependent enzymes are textbook examples of cofactor-aided catalysis, the fine details of their catalytic mech-anism are still lacking (2). Moreover, two types of
intramolec-ular signaling have been observed in ThDP-dependent enzymes. Whereas substrate activation results from communi-cation between an allosteric regulatory site and the active site, the communication among active sites leads to alternating site reactivity (3–5). The molecular mechanism of substrate activa-tion in ThDP-dependent enzymes has been an area of intensive research and debate ever since its first observation in 1967 (6). Phenylpyruvate decarboxylase (PPDC) catalyzes the ThDP-mediated non-oxidative decarboxylation of phenyl-and indolepyruvate to phenyl- or indoleacetaldehyde phenyl-and carbon dioxide (7). In the root-associated bacterium
Azospiril-lum brasilense, AbPPDC catalyzes the second step in the con-version of phenylalanine and tryptophan into phenyl- and indole-acetic acid, respectively (8, 9). The latter compound is a plant hormone, which is responsible for the plant growth pro-moting abilities of A. brasilense (10). AbPPDC shows high structural similarity to other 2-ketoacid decarboxylases from the pyruvate oxidase family (11), such as pyruvate decarboxyl-ase (PDC) and indolepyruvate decarboxyldecarboxyl-ase (IPDC) (12). Sim-ilar to several PDCs, the AbPPDC displays substrate activation with indolepyruvate and other substrates, characterized by sig-moidal v versus [S] plots (7).
Here we present a series of crystal structures of the ThDP-dependent phenylpyruvate decarboxylase in complex with dif-ferent substrates and an analogue of a covalent reaction inter-mediate. These provide new snapshots along the reaction coordinate and unveil the regulatory site and a detailed molec-ular mechanism for allosteric substrate activation.
EXPERIMENTAL PROCEDURES
Protein Expression, Purification, and Preparation of the Complexes—The wild-type AbPPDC was cloned, expressed, and purified as described previously (7, 12). During purification the enzyme activity was monitored with phenylpyruvate as sub-strate using the established coupled optical test with horse liver alcohol dehydrogenase and NADH (13).
Complexes of PPDC with the inhibitors 3-deaza-ThDP 3dThDP) and 2-(1-hydroxyethyl)-3-deaza-ThDP (PPDC-2HE3dThDP) were subsequently obtained by incubating PPDC (purified without addition of ThDP) with 1 mM 3dThDP or
2HE3dThDP for 48 h at 4 °C. Complete exchange of ThDP with the analogues was confirmed by total loss of activity with phenylpyruvate as a substrate. The tertiary complexes with phenylpyruvic acid (PPDC-3dThDP-PPA) and 5-phenyl-2-oxovaleric acid (PPDC-3dThDP-POVA) were then obtained
*This work was supported in part by a research grant from the Fund for Scientific Research-Flanders (FWO-Vlaanderen). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
□S The on-line version of this article (available at http://www.jbc.org) contains supplemental “Results,” Fig. S1, and Tables S1 and S2.
The atomic coordinates and structure factors (code 2Q5J, 2Q5L, 2Q5O, and 2Q5Q) have been deposited in the Protein Data Bank, Research Collabora-tory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).
1Recipient of a post-doctoral grant from the Fund for Scientific
Research-Vlaanderen. To whom correspondence should be addressed: Dept. of Ultrastructure, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Bel-gium. Tel.: 32-2-629-19-49; Fax: 32-2-629-19-63; E-mail: wversees@ vub.ac.be.
2Supported in part by Fund for Scientific Research-Vlaanderen Grant
G.0085.03 and in part by the IAP (IUAP P5/03).
3Supported by a studentship from the Biotechnology and Biological Sciences
Research Council and a Council for Advancement and Support of Educa-tion award from Syngenta.
4The abbreviations used are: ThDP, thiamine diphosphate; PPDC,
phenylpyru-vate decarboxylase; PDC, pyruphenylpyru-vate decarboxylase; IPDC, indolepyruphenylpyru-vate decarboxylase; 3dThDP, 3-deaza-ThDP; 2HE3dThDP, 2-(1-hydroxyethyl)-3-deaza-ThDP; PPA, phenylpyruvic acid; POVA, 5-phenyl-2-oxovaleric acid; POX, pyruvate oxidase.
by incubating the PPDC-3dThDP complex with 100 mM
phenylpyruvate and 5 mM 5-phenyl-2-oxovaleric acid,
respectively.
Crystallization and Data Collection—The PPDC-2HE3dThDP complex was crystallized by the hanging drop vapor diffusion method. Equal volumes of protein solution and precipitant containing 15% polyethylene glycol 4000 (w/v), 10% glycerol (v/v) in 100 mMHepes buffer, pH 7.0, were mixed and
equil-ibrated at 293 K. The crystals were transferred to a cryo-solution containing 20% polyethylene glycol 4000 (w/v), 25% glycerol (v/v) in 100 mM Hepes, pH 7.0, and transferred
immediately to the cryostream. X-ray diffraction data were collected at 100 K to a resolution of 1.85 Å on beamline X11 (EMBL, DESY, Hamburg) using an x-ray wavelength of 0.8157 Å. The 3dThDP, 3dThDP-PPA, and PPDC-3dThDP-POVA complexes were crystallized at 293 K using the hanging drop vapor diffusion method with 11% (w/v) polyeth-ylene glycol 3350 and 0.2Mdi-ammonium tartrate, pH 6.5, as
precipitant solution. Crystals were transferred to a cryo-solu-tion containing 16% (w/v) polyethylene glycol 3350, 0.2M diam-monium tartrate, pH 6.5, 30% glycerol (with addition of PPA and POVA in case of the ternary complexes), and transferred immediately to the cryostream. X-ray diffraction data were col-lected at 100 K to a resolution of 3.2 Å on beamline X11 (EMBL, DESY, Hamburg) using an x-ray wavelength of 0.8162 Å for PPDC-3dThDP, to a resolution of 2.15 Å on beamline X11 (EMBL, DESY, Hamburg) using an x-ray wavelength of 0.8162 Å for PPDC-3DThDP-PPA; and to a resolution of 1.9 Å on beamline BW7A (EMBL, DESY, Hamburg) using an x-ray wavelength of 0.9732 Å for PPDC-3dThDP-POVA.
The diffraction data were indexed and integrated using DENZO and scaled using SCALEPACK (14). Intensities were converted to structure factor amplitudes using TRUNCATE (15). Table 1 sum-marizes the data collection and processing statistics.
Structure Determination and Refinement—Initial phases for all complexes were obtained by molecular replacement with the program PHASER (16) using a monomer of PPDC-ThDP (Pro-tein Data Bank 2NXW) as a search model. The solutions were subjected to the simulated annealing procedure as imple-mented in CNS, and manual model building and inspection of electron density was performed in COOT (17). After several cycles of positional and temperature factor refinement using CNS combined with manual corrections, solvent molecules, cofactors, and alternative conformations were included in the models. Structure refinement was considered complete after crystallographic R-factor and free R-factor had converged, and the difference density was without interpretable features. The final models were checked with the Molprobity web server (18). Refinement statistics are summarized in Table 1. The structural superpositions were performed using the DALI server (19) and/or the program LSQMAN (20). Figures were prepared with PyMOL (21) and Molscript (22).
RESULTS
Substrate and Covalent Intermediate Complexes of PPDC—A standard procedure to obtain detailed structural information on the catalytic mechanism of an enzyme is to solve the struc-ture of slow mutants with trapped substrates or covalent reac-tion intermediates. In a different approach we used unreactive analogues of the coenzyme to trap intermediates in a thiamine-dependent enzyme. 3dThDP is such an unreactive analogue of ThDP, in which the single nitrogen atom of the thiazolium ring is replaced by a carbon (23). Although an excellent steric mimic of the cofactor, 3dThDP electrostatically more closely resem-bles the overall neutral ylide form of thiamine, due to the absence of the positive charge on position 3. AbPPDC in which the naturally occurring ThDP is replaced by 3dThDP proves indeed to be enzymatically inert.
TABLE 1
Data collection and refinement statistics
PPDC-3dThDP PPDC-2HE3dThDP PPDC-3dThDP-PPA PPDC-3dThDP-POVA Data collection Space group C2221 C2221 C2221 C2221 Cell dimensions a, b, c (Å) 100.54, 179.81, 121.03 99.98, 179.05, 120.86 74.90, 145.58, 194.04 74.69, 145.12, 194.11 a, b,␥ (°) 90, 90, 90 90, 90, 90 90, 90, 90 90, 90, 90 Resolution (Å) 50.0-3.2 (3.31-3.20)a 50.0-1.85 (1.92-1.85) 50.0-2.15 (2.23-2.15) 50.0-1.90 (1.97-1.90) Rsym 0.16 (0.65) 0.084 (0.44) 0.11 (0.38) 0.074 (0.48) I/I 11.0 (3.0) 24.1 (3.9) 20.8 (4.7) 18.4 (2.45) Completeness (%) 97.5 (98.3) 98.3 (96.0) 92.8 (83.6) 92.5 (92.0) Redundancy 5.9 (5.9) 7.5 (6.8) 8.4 (8.1) 4.8 (2.5) Refinement Resolution (Å) 50.0-3.2 50.0-1.85 50.0-2.15 50.0-1.90 No. reflections 17,865 90,954 54,182 76,981 Rwork/Rfree 17.61/28.89 17.04/20.43 18.21/23.99 16.32/20.63 No. atoms Protein 7,819 7,943 7,915 7,975 Ligand/ion 54 132 108 132 Water 135 1,010 739 1,117 B-factors Protein 41.4 23.0 39.1 25.1 Ligand/ion 28.8 16.4 44.0 26.0 Water 24.2 35.1 48.4 40.2
Root mean square deviations
Bond lengths (Å) 0.0070 0.0089 0.0090 0.0090
Bond angles (°) 1.313 1.458 1.407 1.430
Ramachandran plot (% in favored, allowed regions)
88.4, 98.3 97.6, 99.5 97.9, 99.8 98.7, 99.7
The crystal structure of PPDC in complex with 3dThDP (PPDC-3dTHDP) was solved to 3.2-Å resolution. This struc-ture is nearly perfectly superimposable on the strucstruc-ture of the native PPDC holo-enzyme (PPDC-ThDP) that we previ-ously solved, showing that replacement of ThDP with 3dThDP causes no structural rearrangements (12) (see sup-plemental materials for all structural details). Because PPDC-3dThDP is enzymatically inert we were able to trap the substrates PPA and POVA in its active site and to solve the structures of these ternary complexes to 2.15-Å (PPDC-3dThDP-PPA) and 1.9-Å (PPDC-3dThDP-POVA) resolu-tion, respectively. Chemical synthesis also allows 3dThDP to be substituted on the C2 to obtain stable analogues of covalent reaction intermediates in the reaction cycle (24). Using such an analogue we solved the structure of PPDC in complex with 2-(1-hydroxyethyl)-3-deaza-ThDP (PPDC-2HE3dThDP) to 1.85-Å resolution to study the structure of the last covalent inter-mediate on the reaction coordinate (Fig. 1, interinter-mediate 5).
All structures show clear electron density with full occu-pancy for all ligands bound to both active sites of the homodimer in the asymmetric unit (Fig. 2). In the PPDC-2HE3dThDP structure (Fig. 2b) the density for the 1-hy-droxyethyl moiety of the intermediate analogue corresponds to a mixture of the R and S enantiomer at the C␣ atom, consistent with the fact that a racemic mixture was used in the co-crystallization. Both enantiomers were modeled with half-occupancy. The overall structure of the active site is very similar in PPDC-3dThDP and PPDC-2HE3dThDP compared with the native holo-enzyme (Fig. 2, a and b). These three structures all exhibit an open active site with the active site loop spanning residues 104 –120, completely dis-ordered (see Verse´es et al. (12) for a detailed description of the PPDC-ThDP active site). In PPDC-3dThDP-PPA and PPDC-3dThDP-POVA, binding of the substrates is accom-panied by a complete structuring of the 104 –120 loop. Upon substrate binding, this long loop folds over the active site bringing His112and His113into the active site pocket (Fig. 2,
cand d). Concomitantly, a second active site loop spanning residues 280 –294 of the neighboring subunit of the tight dimer is reorganized, wedging residues Asp282 and Thr283
deeper into the active site pocket. This structural transition causes a H-bond to be formed between His112 of the first
loop and Asp282of the second loop. The reorganization of
the 280 –294 loop also permits new interactions with the C-terminal helix (e.g. between the side chain of Gln536and
Arg538 and the main chain carbonyls of Asp282 and Ala287
and Ser288, respectively) allowing this helix to bend over
the active site.
A Separate, Regulatory Substrate-binding Site—Apart from the substrate molecules bound in the active sites, the PPDC-3dThDP-PPA and PPDC-3dThDP-POVA structures unambiguously reveal a second site of bound substrate in each subunit. This site is located at a distance of 18 Å from the active site, at the interface of the PYR, R, and PP domains of each subunit. The substrate molecules in this second site are tightly bound by residues coming from the three domains (Fig. 3). In both structures, the carboxyl group of the sub-strate forms charged H-bonds with the Arg60 and Arg215
(bidentate) side chains and an additional H-bond to the main chain amide of Ala397. A third arginine residue, Arg214, is
involved in a cation- stacking interaction with the phenyl moiety of the substrate. The 2-keto oxygen of the substrate forms an H-bond with a protein-bound water molecule. Remarkably, this carbonyl oxygen makes very close (unfa-vorable) contacts with the main chain carbonyls of Met238
(2.9 Å), Arg240(2.8 Å), and Leu395(3.0 Å).
Tertiary Structure Rearrangements upon Substrate Binding— The asymmetric units of all PPDC structures contain a homodimer. In this dimer, each subunit adopts the arche-typical pyruvate oxidase (POX) fold consisting of the PYR, R, and PP domains (25). The secondary complexes PPDC-ThDP, PPDC-3dPPDC-ThDP, and PPDC-2HE3dThDP are very similar in subunit architecture (see supplemental Table 1S for a full list of root mean square deviations upon superpo-sition). Upon substrate binding in the active and regulatory sites, a change in this subunit architecture involving a domain rotation is observed in both ternary complexes (PPDC-3dThDP-PPA and PPDC-3dThDP-POVA). When the PYR and PP domains of PPDC-ThDP and either of the substrate complexes are superimposed, the R domains differ by a rotation of about 12° (Fig. 4). This domain rotation is accompanied by a large rearrangement of active site loop 280 –294 of the R domain and the concomitant ordering of active site loop 104 –120 of the PYR domain of the neighbor-ing subunit causneighbor-ing the complete closure of the active sites in PPDC-3dThDP-PPA and PPDC-3dThDP-POVA.
Quaternary Structure Rearrangements upon Substrate Binding—PPDC adopts a homotetrameric assembly in solution (7). In all the crystal structures, the two tight dimers constitut-ing the biological tetramers are related through a 2-fold crys-tallographic symmetry axis (7, 12). However, large differences exist between the tetramer architecture of the binary
complexes (PPDC-ThDP, PPDC-3dThDP, and
PPDC-2HE3dThDP) and the ternary complexes with substrates bound at the active and the regulatory sites (PPDC-3dThDP-PPA and PPDC-3dThDP-POVA). In the binary complexes, the non-perpendicular arrangement of non-crystallographic axes relating the monomers in the asymmetric unit and the crystallographic axis relating the two dimers results in an asymmetrical tetramer assembly, best described as an asym-metrical dimer of dimers (Fig. 5a). For the ternary complexes the non-crystallographic symmetry axes relating the two subunits in the dimer intersect with the crystallographic axis at an angle of 90°, resulting in a dimer of dimers with pseudo 222 symmetry (Fig. 5b). In going from the asymmetrical dimer of dimers observed for the binary complexes to the symmetrical dimer of dimers of the ternary complexes, one dimer has to be rotated by about 34°, vis a` vis to the second dimer. This also has implications for the dimer-dimer inter-faces. Whereas the AC and BD interfaces are different in the asymmetrical tetramers (see Fig. 5, for subunit nomencla-ture) these interfaces are the same for the symmetrical tetramers.
A Cascade of Conformational Changes Links the Regulatory Substrate-binding Site to the Active Site—Comparison of the binary complexes (ThDP, 3dThDP, and
PPDC-2HE3dThDP) with the ternary com-plexes (PPDC-3dThDP-PPA and PPDC-3dThDP-POVA) uncovers a series of coupled local conforma-tional changes running from the regulatory substrate-binding site to the active site (see Fig. 6). Binding of a substrate molecule in the regula-tory site causes a change in side chain conformation of Arg214
lead-ing to the formation of a cation- interaction with the phenyl group of the substrate. The void left by the arginine side chain gets occupied by the side chain of Leu242. This
rear-rangement of Leu242, together with
the steric repulsion caused by the binding of the 2-ketoacid moiety of the substrate and the loss of the H-bond between Arg214 and the
main chain carbonyl of Arg240,
induces a relocation of the entire region between Phe237 and Pro247.
The changes in the Phe237–Pro247
segment lead to further structural rearrangements in two distinct branches both leading to the active site. In a first branch, changes at Met238force the Tyr400side chain to
flip around with a concomitant dis-placement of its hydroxyl group by 9.8 Å. This conformational change in its turn forces a relocation of Leu109 of the neighboring subunit
and the associated structuring of the entire 104 –120 loop, bringing the two residues His112and His113into
the active site. The second branch of the cascade is initiated by the steric clash of Phe237with Phe285located
on the 280 –294 active site loop. This causes a rearrangement of this entire loop, the most obvious change being a 13-Å displacement of the Phe285side chain.
Reorgani-zation of the 280 –294 loop (branch 2) liberates the necessary space for the 104 –120 loop (branch 1) to enter the active site. Asp282is also
shifted into the active site in this process (a shift of 5 Å) and now forms a H-bond with His112. This
cascade of transmitted conforma-tional changes clearly provides a link between substrate binding in the regulatory substrate-binding site and structural changes in the active site.
DISCUSSION
Snapshots along the Catalytic Cycle of a ThDP-dependent Decarboxylase—Multiple studies have been reported dealing with the complex kinetics of ThDP-dependent enzymes. How-ever, a full interpretation of this wealth of data has always been hampered by a lack of structures visualizing the relevant reac-tion intermediates. Only recently structures of some of these intermediates have been solved: the lactyl-ThDP intermediate in the active site of pyruvate oxidase (intermediate 3, see Fig. 1); the planar enamine intermediate in the active site of pyruvate oxidase and transketolase (intermediate 4B, see Fig. 1) and the non-planar, more carbanion-like form of the latter intermedi-ate in the active site of branched-chain 2-ketoacid dehydrogen-ase (intermediate 4A, see Fig. 1) (26 –28). In the current paper
we present the crystal structures of two other reaction intermediates. The use of the non-reactive cofactor analogue 3-deaza-ThDP allowed to solve structures in complex with the genuine substrates phenylpyruvic acid and 5-phenyl-2-oxovaleric acid to high resolution, providing images of the enzyme-substrate Michaelis complex for this class of enzymes (intermediate 2, see Fig. 1). More-over, the structure of PPDC in com-plex with 2-(1-hydroxyethyl)-3-deaza-ThDP provides a structural model of the last tetrahedral 2-(1-hydroxyethyl)-ThDP intermediate on the reaction cycle (intermediate 5, see Fig. 1). Following our results, snapshots of five of six intermedi-ates along the reaction cycle are now available, albeit from different but related ThDP-dependent enzymes (intermediates 1–5 on Fig. 1). Assuming that the geometry of the intermediates is conserved among ThDP enzymes, educated modeling of the 2-(3-phenylactyl)-ThDP and 1-hydroxy-2-phenylethylenamine intermediates in the active site of PPDC enables us to follow the decarboxylation reaction of phe-nylpyruvate step by step in a single active site (Fig. 7).
It is well established that the first step in the reaction cycle of the ThDP-dependent decarboxylases involves activation of the cofactor by abstracting the proton of the C2 of the thiazolium ring (29). Recent studies have shown that a conserved H-bond between a glutamate (Glu48 in PPDC) and the N1⬘ of the
amino-pyrimidine group of the cofactor stabilizes the imino tautomer at the N4⬘, allowing this group to cycle between the amino and imino forms (30, 31). Similar to other ThDP-dependent enzymes, the structures of PPDC-ThDP and PPDC-3dThDP show that the induced V conformation of the bound cofactor brings the N4⬘ in close proximity to the C2 allowing the former group in its imino form to act as a general base in the deproto-nation of C2 (Fig. 7a).
Upon binding of the substrates PPA or POVA, the active site loops 104 –120 and 280 –294 and the C-terminal helix fold over the active site (Fig. 7b). These structures show the carbonyl
FIGURE 1. General reaction scheme for the decarboxylation of 2-keto acids by PPDC. B(H) represent the general acids/bases required for the (de)-protonation steps. Abstraction of the proton on the C2 carbon by the N4⬘ amino/imino group of the aminopyrimidine ring of ThDP (29, 31) results in an ylide negatively charged on C2 (intermediate 1). This C2 carbanion subsequently attacks a carbonyl carbon of a bound substrate molecule (intermediate 2), giving in the case of phenylpyruvate as a substrate the tetrahedral 2-(3-phenyl-lactyl)-ThDP intermediate (intermediate 3). Bond cleavage with expulsion of carbon dioxide leads to a carbanion (interme-diate 4A), which is resonance stabilized by its enamine form (interme(interme-diate 4B). This then reacts with an electrophile, which in the case of PPDC is simply a proton, giving 2-(1-hydroxy-2-phenyl-ethyl)-ThDP (intermediate 5). Finally deprotonation of the␣-hydroxylgroupandcleavageofthecarbon-carbonbondbetweenC2ofThDPand C␣ of the intermediate leads to release of the aldehyde product (intermediate 6), with regeneration of the ylide.
FIGURE 2. Close up view of the active sites of PPDC-3dThDP (a), PPDC-2HE3dThDP (b), PPDC-3dThDP-PPA (c), and PPDC-3dThDP-POVA (d). Residues provided by different subunits are color-coded green and
cyan. Parts of the 104 –120 and 280 –294 loops, and the C-terminal helix are shown in a coil representation.
Cofactor and intermediate analogues are colored yellow, and substrates magenta. The Mg2⫹ion and a protein bound water molecule are represented by a gray and red sphere, respectively. Interactions with the substrate and the water molecule are indicated by orange dotted lines. The electron density (contoured at 5 for PPDC-2HE3dThDP and PPDC-3dThDP-POVA, and 4 for PPDC-3dThDP and PPDC-3dThDP-PPA) of a Fo⫺ Fc simu-lated annealed omit map calcusimu-lated without cofactors and substrates is shown (blue mesh).
carbon (C␣) of the substrates at 3.4 Å from the C2, ideally posi-tioned for nucleophilic attack by the C2 carbanion. The carbox-ylate of the substrate is held in place by interactions with the main chain amide of Asp25, by His113, and by an unusual short
interaction with an enzyme-bound water molecule (2.3 Å). This water molecule is conserved in the active sites of Zymomonas
mobilisPDC (32), oxalyl-CoA decarboxylase (33), and pyruvate oxidase (26).
The next step in catalysis, the carbon-carbon bond formation between the C2 of ThDP and the carbonyl carbon of the substrate, requires a steric orientation of the substrate and a protonation of its carbonyl oxygen. Modeling of the 2-(3-phenyllactyl)-ThDP intermediate in the active site of PPDC (starting from the structure of 2-lactyl-ThDP bound to the active site of Lactobacillus plantarum POX (26)) shows the C2␣ hydroxyl group within interaction distance to His113and the
N4⬘ cofactor, now in its acidic amino tautomeric form (Fig. 7c).
By analogy to the current view for other ThDP enzymes we propose the 4⬘-NH2 as the general acid in
carbonyl protonation (34 –36). Carbon-carbon bond cleavage in the substrate with release of CO2gives the following covalent ␣-carbanion/enamine intermedi-ate, which was modeled as the enamine in the active site of PPDC (starting from the structure of the enamine bound to the active site of
L. plantarum POX (26); Fig. 7d). Protonation of this intermediate at C␣ will yield the last tetrahedral intermediate, 2-(1-hydroxy-2-phenylethyl)-ThDP. The protona-tion of this sp2hybridized enamine
C␣ could in principle occur from either side (si or re face) of the prochiral center, thus determining the stereochemistry of the next intermediate. The structure of PPDC-2HE3dThDP does not resolve this issue because it has both enantiomers of the 2-hydroxyethyl moiety bound in its active site. This is probably a consequence of the presence of a small, non-physiological methyl side chain on 2HE3ThDP. Indeed, modeling of a phenyl group onto the-methyl of 2HE3dThDP shows that only the R enantiomer of 2-(1-hydroxy-2-phenylethyl)-ThDP can be accommodated in the active site without inducing significant steric clashes with the protein (Fig. 7e). This provides indirect but strong evidence that addition of the proton to the C␣ of the enamine occurs from the si face of the pro-chiral center, excluding the 4⬘-amino/imino group as the catalytic acid in this step. In PDC and IPDC a Glu-Asp-His catalytic triad has been implicated in protonation of the enamine (Glu468-Asp29-His115
in Enterobacter cloacae IPDC numbering) (35). However, in PPDC the glutamate residue of this triad is replaced by a leucine (Leu462) (12). In PPDC, Asp25(equivalent to Asp29in IPDC) is
part of an Asp25-His112-Asp282triad and could fulfill the role of
general acid, although it is located at about 4.9 Å from the C␣. Alternatively the active site water could fulfill the role of proton donor. This water molecule is located at 4 Å from the C␣ of the product 2-(1-hydroxy-2-phenylethyl)-ThDP. A similar role has been proposed for the corresponding water molecule in oxalyl-CoA decarboxylase (33). In this scenario the Asp25-His112
-Asp282triad could be used to activate the water molecule by
shuttling a proton into the active site. The latter proposal comes with a caveat, because activation of the water molecule requires an intact catalytic triad and hence fully structured active site loops. However, one would expect the active site loops to be open if CO2has diffused out from the active site.
Activation of the water molecule in the active site by the Asp-His-Asp triad thus requires the decarboxylation and protona-tion of the carbanion/enamine to occur quasi simultaneously as recently proposed by Kluger (37). From this it would also follow that the carbanion/enamine intermediate maintains significant carbanion character, with tetrahedral geometry, throughout this process.
FIGURE 3. The regulatory substrate-binding site of PPDC-3dThDP-POVA: location of the regulatory sub-strate-binding site at the juncture of the three domains in a PPDC subunit (a), and close up view of the interactions with the regulatory substrate (b). The PYR, R, and PP domains in a are color-coded blue, green, and red, respectively, and the same coloring is maintained in b for the C-atoms of the amino acid residues in the regulatory site. The POVA molecules bound in the active site and in the regulatory site are shown as sticks and colored magenta and orange, respectively. In panel b the electron density (contoured at 5) of a Fo⫺ Fc simulated annealed omit map calculated around the regulatory POVA molecule is shown as a blue mesh.
FIGURE 4. Tertiary structure rearrangements upon substrate binding. A superposition of the binary complex PPDC-ThDP and the ternary complex PPDC-3dThDP-POVA is shown, zoomed in on one subunit. Whereas the PYR and PP domains of PPDC-ThDP and PPDC-3dThDP-POVA superimpose nearly perfectly, the R domain rotates by about 12° upon substrate binding. PPDC-ThDP is shown in gray; PPDC-3dPPDC-ThDP-POVA is colored according to the domain, with the PYR, R, PP domains, and the intervening loops colored blue,
green, red, and yellow, respectively. The mobile 280 –294 loop of the R domain
and the 104 –120 loop of the PYR domain of the neighboring subunit are colored orange and are indicated by arrows. His112and His113of loop 104 –120
In the last step of the reaction cycle the C␣-hydroxyl group of 2-(1-hydroxy-2-phenylethyl)-ThDP is deprotonated with concomitant cleavage of the C2-C␣ covalent bond and release of the second product phenylacetaldehyde. The PPDC-2HE3dThDP structure shows the hydroxyl group within 2.5 Å from the N4⬘ imino group (Fig. 7e), confirming the current view that this group acts as a proton acceptor in the product release step (34).
In conclusion, these structures corroborate the central role of the N4⬘ amino/imino group of the cofactor in three of the four proton transfer steps of ThDP-catalyzed decarboxylation reactions (30, 31, 34). These pivotal protonation/deprotonation steps are promoted by a first proton relay involving the amin-opyrimidine ring and Glu48. Protonation of the
enamine/car-banion intermediate is mediated by a second Asp25-His112-Asp282
pro-ton relay, either direct through Asp25 or via a tightly bound water
molecule.
Allosteric Substrate Activation in PPDC—Nearly all pyruvate decar-boxylases studied to date (with the exception of some bacterial PDC’s like the Z. mobilis PDC) are sub-jected to substrate activation, char-acterized by sigmoidal v/[S] plots and time-dependent slow activation resulting in a marked lag phase in product formation. Ever since the first observation of homotropic allosteric activation in the PDC from wheat germ in 1967 (6), the elucidation of the molecular mech-anism behind this phenomenon has been an active area of research. The currently accepted kinetic model for this cooperative behavior pre-dicts that, only upon binding of sub-strate in a regulatory site, a slow activation process converts (nearly) inactive enzyme into the active form (1). Many different theories have been proposed throughout the years to explain the molecular mecha-nism behind this behavior (38, 39). In Saccharomyces cerevisiae PDC,
the most commonly accepted
mechanism for substrate activation involves a cysteine (Cys221) as the
site of covalent binding of the regu-latory substrate molecule (see sup-plemental “Results” and Fig. S1) (38, 40). A full understanding of any of these mechanisms is missing, how-ever, mainly due to the lack of struc-tures of substrate complexes with an occupied regulatory site. Here, we used the inactive cofactor analogue 3dThDP to obtain structures of PPDC in complex with its sub-strates PPA and POVA. Whereas PPDC shows clear positive cooperativity with POVA as a substrate (Hill coefficient of 1.93), the deviation from Michaelis-Menten kinetics is rather small for PPA (Hill coefficient of 1.26 when fitted on the Hill equation), and PPDC was reported to behave with Michaelis-Menten kinetics toward PPA (7). Each subunit reveals two sep-arate sites with substrate bound, corresponding to the active site and a regulatory site. The regulatory site is located at 18 Å from the active site, at the interface of the three domains of the monomer (Fig. 3).
Comparison of the substrate-bound ternary PPDC com-plexes (PPDC-3dThDP-PPA and PPDC-3dThDP-POVA) with the binary complexes (PPDC-ThDP, PPDC-3dThDP, and
FIGURE 5. Quaternary structure rearrangements upon substrate binding. The biological tetramers of the binary complex PPDC-2HE3dThDP (a) and the ternary complex PPDC-3dThDP-POVA (b) are depicted with their AB homodimer in the same orientation. The transition from an asymmetrical dimer of dimers in substrate-free PPDC to a symmetrical dimer of dimers in substrate-bound PPDC is accompanied by a 34° rotation of the CD dimer vis a` vis the AB dimer. Cofactors and substrates are shown as sticks with the carbon atoms of 3dThDP and 2HE3dTHDP colored yellow and the POVA substrates colored purple. The crystallographic 2-fold axis relating the two dimers is indicated in orange. The two NCS-axes relating the subunits in the asymmetric unit are indicated as blue and red bars.
FIGURE 6. Intramolecular signal transduction via a cascade of conformational changes. A superposition of the relevant residues of the unactivated binary complex PPDC-ThDP (in gray) with the activated ternary com-plex PPDC-3dThDP-POVA (in green and blue for residues of the A and B subunit, respectively) is shown. POVA molecules in the regulatory and active site are shown in magenta. Binding of a substrate molecule in the regulatory site causes a rearrangement of Arg214, with concomitant rearrangement of the region between
Phe237and Leu242. These rearrangements are transmitted via Tyr400and Phe285to the 104 –120 and 280 –294
loops that fold over the active site. A H-bond between Asp282and His112is hereby formed completing the
proposed Asp25-His112-Asp282catalytic triad. This sterical relay mechanism hence provides a clear link between
PPDC-2HE3dThDP) reveals significant changes in the protein conformation that are linked to the binding of the substrate in the regulatory site. These structural changes are taking place on three different levels and convert the unactivated enzyme into the substrate-bound activated form. 1) Fig. 5 illustrates the quaternary structural transition upon binding of a substrate in the regulatory site of PPDC. A rotation of 34° of one dimer relative to the other converts the PPDC tetramer from an ametric dimer of dimers in the unactivated state to a fully sym-metrical dimer of dimers in the substrate-bound activated state.
Significant changes at the quaternary level were also observed in S. cerevisiae PDC upon binding of the synthetic activator pyruvamide (39). Remarkably, this enzyme undergoes the opposite transition: from a symmetrical tetramer in the unbound state to an asymmetrical tetramer in the pyruvam-ide-bound state. 2) Fig. 4 illustrates the changes occurring at the tertiary level. The subunits of the unactivated and activated enzymes differ by a 12° rotation of the R domain relative to the PYR and PP domains. This rotation can be triggered by local rearrangements in the R domain upon binding of a substrate
FIGURE 7. Snapshots along the reaction cycle of phenylpyruvate decarboxylation catalyzed by PPDC. Each panel shows a different intermediate as also defined in the legend to Fig. 1. a, free enzyme (or ylide); b, Michaelis complex as provided by the structure of PPDC-3dThDP-PPA; c, modeled structure of the PPDC-2-(3-phenyllactyl)-ThDP intermediate; d, modeled structure of the enamine intermediate; e, PPDC-2-(1-hydroxy-2-phenylethyl)-3dThDP intermediate as obtained by modeling a phenyl side group in the structure of PPDC-2HE3dThDP; f, the phenylacetaldehyde product complex for which no structure is available yet. Amino acids colored cyan and green are contributed from neighboring subunits. The carbon atoms of the ThDP (or 3dThDP) moiety of the intermediates are shown in yellow, whereas carbon atoms stemming from the phenylpyruvate substrate are colored magenta. The Mg2⫹ion and an active site-bound water
molecule at the regulatory site located between the PYR, R, and PP domains. These rearrangements in the R domain involve a segment spanning residues Phe237to Pro247and the Arg214side
chain. In the unactivated state Arg214 is hydrogen bonded
within the same domain to Arg240. In the substrate-bound
enzymes Arg214forms other H-bonds with the main chain
car-bonyls of Tyr61and Arg60, located on the PYR domain. The
rotation of the R domain is also accompanied by the movement of the active site loop 280 –294 and the concomitant structur-ing of the interactstructur-ing 104 –120 loop of the PYR domain of the neighboring subunit in a tight dimer. 3) Comparison of the activated and unactivated structures in Fig. 6 also reveals a sig-nal transduction route that communicates structural changes at the regulatory site to the active site via a cascade of sterical clashes transmitted from one amino acid residue to the next. Substrate binding at the regulatory site changes the side chain conformation of Arg214and triggers this cascade. The signal is
then transmitted via Arg214, Phe237, Met238, and Tyr400to two
distinct active site loops spanning residues 104 –120 and 280 – 294 (see “Results”). These loops fold over the active site hence providing catalytic residues His112and His113(41, 42). A new
H-bond between Asp282of loop 280 –294 and His112of loop
104 –120 completes the Asp25-His112-Asp282catalytic triad,
which we propose to play a role in protonation of the enam-ine/carbanion intermediate.
These results thus provide the first pictures of a molecular mechanism underlying the kinetically observed allosteric substrate activation behavior of ThDP-dependent enzymes. Indeed, the observed structural changes between the acti-vated ternary complexes and the unactiacti-vated binary com-plexes link the binding of a substrate molecule in a separate regulatory site with loop closures and concomitant align-ment of the catalytic machinery in the active site. Once sub-strate is bound in the regulatory sites, multiple turnovers might occur in enzymatically competent active sites with facilitated, fast loop opening and closing, explaining the often observed slow time-dependent activation of ThDP-de-pendent enzymes. In such a hysteretic model for positive cooperativity the observed Hill coefficient will depend on the affinity of the regulatory site for the substrate (43). Very high affinity binding of a substrate in the regulatory site will pull the enzyme completely toward the activated form for all substrate concentrations used for obtaining the v versus [S] curve. This will yield nearly Michaelis-Menten kinetics as observed for PPA as a substrate. Lower affinity binding to the regulatory site will cause significant deviations from hyper-bolicity as observed for POVA.
A Versatile PYR-R-PP Domain Interface—A clue to the evo-lution of the signal transduction pathway in PPDC comes from comparison with POX. POX is a homologous ThDP-dependent enzyme that, in the presence of phosphate, catalyzes the oxida-tive decarboxylation of pyruvate with formation of carbon diox-ide, hydrogen peroxdiox-ide, and acetylphosphate (26). The ThDP-dependent POX also uses a flavin adenine dinucleotide cofactor in its reaction mechanism (25). This FAD is bound in a deep cleft, with its ADP moiety located on the R domain and the FMN part at the interface of the PYR, R, and PP domains. Superposition reveals that the residues involved in the signal
transduction pathway of PPDC coincide with the FMN binding groove in POX. More surprisingly, the activated form of PPDC provides a cavity almost large enough to accommodate the FMN moiety of FAD. In the unactivated form of PPDC the large conformational changes of the residues in the signal transduc-tion pathway cause these residues to occupy this cavity. This observation suggests that the interface between the PYR, R, and PP domains provides a versatile scaffold adaptable to serve dif-ferent purposes during evolution. In POX, this interface evolved toward a binding pocket for an additional cofactor. In PPDC this interface has evolved to a flexible region adapted to dynamical intramolecular signal transduction.
Acknowledgment—We acknowledge the EMBL for the use of synchro-tron beam time at the Hamburg outstation (DESY, Hamburg, Germany).
Note Added in Proof—While this manuscript was under consider-ation for publicconsider-ation, a study on oxalyl-CoA decarboxylase was pub-lished that also used 3-deaza-ThDP to allow a crystal structure of an enzyme-substrate complex to be obtained (Berthold, C. L., Toyota, C. G., Moussatche, P., Wood, M. D., Leeper, F., Richards, N. G. J., and Lindqvist, Y. (2007) Structure 75, 853– 861).
REFERENCES
1. Schowen, R. L. (1998) in Comprehensive Biological Catalysis, A
Mechanis-tic Reference(Sinnott, M., ed) Vol. II, pp. 217–266, Academic Press, San Diego
2. Jordan, F. (2003) Nat. Prod. Rep. 20, 184 –201
3. Sergienko, E. A., and Jordan, F. (2002) Biochemistry 41, 3952–3967 4. Jordan, F., Nemeria, N., and Sergienko, E. A. (2005) Acc. Chem. Res. 38,
755–763
5. Frank, R. A., Titman, C. M., Pratap, J. V., Luisi, B. F., and Perham, R. N. (2004) Science 306, 872– 876
6. Davies, D. D. (1967) Biochem. J. 104, 50P
7. Spaepen, S., Verse´es, W., Gocke, D., Pohl, M., Steyaert, J., and Vanderley-den, J. (2007) J. Bacteriol. 189, 7626 –7633
8. Somers, E., Ptacek, D., Gysegom, P., Srinivasan, M., and Vanderleyden, J. (2005) Appl. Environ. Microb. 71, 1803–1810
9. Costacurta, A., Keijers, V., and Vanderleyden, J. (1994) Mol. Gen. Genet. 243,463– 472
10. Dobbelaere, S., Croonenborghs, A., Thys, A., Vande Broek, A., and Vanderleyden, J. (1999) Plant Soil 212, 155–164
11. Duggleby, R. G. (2006) Acc. Chem. Res. 39, 550 –557
12. Verse´es, W., Spaepen, S., Vanderleyden, J., and Steyaert, J. (2007) FEBS J. 274,2363–2375
13. Weiss, P. M., Garcia, G. A., Kenyon, G. L., Cleland, W. W., and Cook, P. F. (1988) Biochemistry 27, 2197–2205
14. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307–326
15. French, S., and Wilson, K. (1978) Acta Crystallogr. Sect. A 34, 517–525 16. McCoy, A. J., Grosse-Kunstleve, R. W., Storoni, L. C., and Read, R. J. (2005)
Acta Crystallogr. Sect. D Biol. Crystallogr. 61,458 – 464
17. Emsley, P., and Cowtan, K. (2004) Acta Crystallogr. D Biol. Crystallogr. 60, 2126 –2132
18. Lovell, S. C., Davis, I. W., Adrendall, W. B., de Bakker, P. I., Word, J. M., Prisant, M. G., Richardson, J. S., and Richardson, D. C. (2003) Proteins 50, 437– 450
19. Holm, L., and Sander, C. (1998) Nucleic Acids Res. 26, 316 –319 20. Collaborative Computational Project Number 4 (1994) Acta Crystallogr.
D Biol. Crystallogr. 50,760 –763
21. DeLano, W. L. (2002) The PyMOL Molecular Graphics System, DeLano Scientific, San Carlos, CA
23. Mann, S., Perez Melero, C., Hawksley, D., and Leeper, F. J. (2004) Org.
Biomol. Chem. 2,1732–1741
24. Leeper, F. J., Hawksley, D., Mann, S., Perez Melero, C., and Wood, M. D. H. (2005) Biochem. Soc. Trans. 33, 772–775
25. Muller, Y. A., and Schulz, G. E. (1993) Science 259, 965–967
26. Wille, G., Meyer, D., Steinmetz, A., Hinze, E., Golbik, R., and Tittmann, K. (2006) Nat. Chem. Biol. 2, 324 –328
27. Fiedler, E., Thorell, S., Sandalova, T., Golbik, R., Konig, S., and Schneider, G. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, 591–595
28. Machius, M., Wynn, R. M., Chuang, J. L., Li, J., Kluger, R., Yu, D., Tomchick, D. R., Brautigam, C. A., and Chuang, D. T. (2006) Structure 14, 287–298 29. Breslow, R. (1958) J. Am. Chem. Soc. 80, 3719 –3726
30. Jordan, F., Nemeria, N. S., Zhang, S., Yan, Y., Arjunan, P., and Furey, W. (2003) J. Am. Chem. Soc. 125, 12732–12738
31. Kern, D., Kern, G., Neef, H., Tittmann, K., Killenberg-Jabs, M., Wikner, C., Schneider, G., and Hu¨bner, G. (1997) Science 275, 67–70
32. Dobritzsch, D., Ko¨nig, S., Schneider, G., and Lu, G. (1998) J. Biol. Chem. 273,20196 –20204
33. Berthold, C. L., Moussatche, P., Richards, N. G., and Lindqvist, Y. (2005)
J. Biol. Chem. 280,41645– 41654
34. Nemeria, N., Baykal, A., Joseph, E., Zhang, S., Yan, Y., Furey, W., and Jordan, F. (2004) Biochemistry 43, 6565– 6575
35. Schu¨tz, A., Golbik, R., Ko¨nig, S., Hu¨bner, G., and Tittmann, K. (2005)
Biochemistry 44,6164 – 6179
36. Nemeria, N., Chakraborty, S., Baykal, A., Korochkina, L. G., Patel, M. S., and Jordan, F. (2007) Proc. Natl. Acad. Sci. U. S. A. 104, 78 – 82 37. Kluger, R., Ikeda, G., Hu, Q., Cao, P., and Drewry, J. (2006) J. Am. Chem.
Soc. 128,15856 –15864
38. Wang, J., Golbik, R., Seliger, B., Spinka, M., Tittmann, K., Hu¨bner, G., and Jordan, F. (2001) Biochemistry 40, 1755–1763
39. Lu, G., Dobritzsch, D., Baumann, S., Schneider, G., and Ko¨nig, S. (2000)
Eur. J. Biochem. 267,861– 868
40. Baburina, I., Li, H., Bennion, B., Furey, W., and Jordan, F. (1998)
Biochem-istry 37,1235–1244
41. Schenk, G., Leeper, F. J., England, R., Nixon, P. F., and Duggleby, R. G. (1997) Eur. J. Biochem. 248, 63–71
42. Joseph, E., Wei, W., Tittmann, K., and Jordan, F. (2006) Biochemistry 45, 13517–13527
43. Krieger, F., Spinka, M., Golbik, R., Hu¨bner, G., and Ko¨nig, S. (2002) Eur.
