Logo of jbcAbout JBCASBMBSubmissionsSubscriptionsContactJBCThis Article
J Biol Chem. 2009 Oct 2; 284(40): 27037–27041.
Published online 2009 Jul 27. doi:  10.1074/jbc.R109.040568
PMCID: PMC2785633

Structural Insights into the Mechanism of Phosphoenolpyruvate Carboxykinase Catalysis*

Mammalian PEPCK2 catalyzes the reversible formation of PEP from OAA and GTP (or ITP) in a divalent cation-dependent reaction (Scheme 1), as was elegantly discussed in the first minireview of this series on PEPCK (1).

PEPCK-catalyzed interconversion of OAA and PEP.

In this third minireview, high-resolution crystal structures of mammalian PEPCK are examined to gain insights into the mechanism of PEPCK catalysis, including the reaction's reversibility and nucleotide specificity. Regarding reaction reversibility, PEPCK is responsible for regeneration of the high-energy phosphoryl donor PEP from the unstable, activated β-keto acid OAA. When coupled with pyruvate carboxylase, PEPCK reverses the essentially irreversible formation of pyruvate and ATP from PEP and ADP in the glycolytic reaction catalyzed by pyruvate kinase. As illustrated (Fig. 1), PEPCK could achieve this feat by stabilizing the inherently unstable enolate form of pyruvate generated by decarboxylation of OAA (Scheme 1). Stabilization of this intermediate would reduce the energetic cost for phosphoryl transfer by ∼30 kJ mol−1 relative to direct reversal of the pyruvate kinase-catalyzed reaction. An energetic driving force for the pyruvate kinase reaction is the favorable tautomerization of the high-energy enol to its corresponding keto form; in contrast, by stabilizing the enolate, PEPCK could prevent its energetically favorable protonation and tautomerization, allowing phosphoryl transfer to occur. Thus, by stabilizing this intermediate in a high-energy state, the PEPCK reaction would be energetically rendered freely reversible; the crystal structures that will be described indicate that PEPCK does, in fact, stabilize the enolate intermediate.

Diagram representing the reaction coordinates for the pyruvate kinase-catalyzed (A) and PEPCK-catalyzed (B) reactions. The standard free energy values given are approximate values based upon the average values from a number of literature sources. The ...

The recent structures of PEPCK from human, rat, and chicken (25), the enzymes from Trypanosoma cruzi (6), Anaerobiospirillum succiniciproducens (7), and Corynebacterium glutamicum (8), and earlier work on the isozyme from Escherichia coli (913) illustrate that the active-site residues and architecture are well conserved, despite what is rather poor overall sequence homology when comparing members of the ATP- and GTP-dependent families.3 As detailed in this minireview, the cationic environment of the active site, dominated by the juxtaposition of two divalent metal ions and the positioning of lysine and arginine residues, is well suited to allow for the stabilization of the enolate intermediate discussed above and to facilitate phosphoryl transfer.

An informative aspect of the PEPCK-catalyzed reaction revealed by the recent structural data on the GTP-dependent isozyme from rat is the illumination of the previously unappreciated role of conformational changes occurring in the active site during the catalytic cycle (5). The most prevalent mobile feature illustrated by the structural work is a 10-residue Ω-loop lid domain whose closure is potentially capable of protecting the enolate intermediate (Fig. 2) (25). A similar domain is present in ATP-dependent PEPCK, as represented by the E. coli enzyme, which was the first PEPCK to be structurally characterized (9). The structural data on PEPCK demonstrate that only upon closure of the lid domain are the substrates positioned correctly for catalysis to occur (5). Furthermore, another loop domain, the ubiquitous P-loop or kinase-1a motif in the GTP-dependent PEPCKs, also shows dynamic behavior, adapting various conformations correlated with substrate binding. The potential role of the dynamic P-loop in catalysis is of interest because it contains a reactive cysteine residue that is conserved in all GTP-dependent PEPCKs and whose specific modification has been known for 2 decades to result in the inactivation of the enzyme (14, 15). As described below, recent structural work characterizing the low-energy conformational states that define the reaction coordinate of the enzyme-catalyzed reaction (25, 16), considered together with previous biochemical studies, has allowed a relatively detailed picture of the mechanism of catalysis utilized by PEPCK to emerge. Both the role of the positively charged active site and the important conformational changes occurring within that site are discussed in the context of an integrated mechanism for PEPCK-mediated catalysis.

Crystallographic images defining the chemical reaction path of PEPCK-mediated conversion of OAA to PEP. A schematic drawing to aid in the interpretation of the structural data is presented on the right-hand side of each panel. In the left-hand images, ...

Structural Snapshots of a Stepwise Mechanism for PEPCK

Formation of the Michaelis Complex

The crystallographic structures of PEPCK in complex with either OAA or its analog 3-sulfopyruvate demonstrate that OAA binds to the enzyme by directly coordinating to the M14 manganese ion with its C-35 and C-4 carbonyl oxygens displacing two of three water molecules coordinated to that cation in the structure of PEPCK-Mn2+ (Fig. 2A). Structure-function studies on rat cytosolic PEPCK indicate that the cis-planar arrangement of metal-coordinating atoms is essential in substrate/inhibitor recognition by PEPCK, which explains the poor inhibition observed in prior kinetic studies by OAA analogs lacking this feature (1618). In addition to the favorable interactions between OAA and the M1 Mn2+ ion, binding of OAA is facilitated through interactions with Arg-876 and Ser-286 (19). GTP binds to PEPCK within the unique nucleotide-binding site of GTP-dependent PEPCKs (Fig. 3), with β- and γ-phosphoryl groups coordinating to the M2 metal (Fig. 2A). These two coordinating atoms from GTP, in conjunction with the O-γ of the P-loop Thr-291 side chain and three water molecules, complete the octahedral coordination geometry of the M2 metal ion. Upon GTP binding to the PEPCK·OAA complex, forming the Michaelis complex, the remaining water molecule coordinated to the M1 metal ion is displaced, fulfilling its octahedral coordination geometry. This coordination of the γ-phosphate allows it to act as a bridging ligand between M1 and M2 metal ions (Fig. 2, A–C). Furthermore, coordination of OAA and GTP to the M1 and M2 metal ions places the C-3 carbonyl oxygen of OAA and the γ-phosphate of GTP in the perfect orientation, albeit at a distance that is too great (see “Activation of Catalysis upon Lid Closure”), for direct in-line phosphoryl transfer, consistent with the observed stereochemistry of the reaction (20, 21). Thus, in the fully ligated state, the M1 metal is octahedrally coordinated by three protein ligands (Lys-244, His-264, and Asp-311) and three oxygen atoms of the OAA and GTP substrates. In a role consistent with its known function in kinases, the P-loop lysine (Lys-290) bridges the β- and γ-phosphoryl groups, facilitating the leaving group ability of the γ-phosphate (22). Furthermore, the importance of Arg-405 in forming a bidentate salt bridge with the phosphoryl group being transferred is illustrated in the complexes representing the two Michaelis states for the reversible reaction (Fig. 2, B and D). The bidentate interaction of Arg-405 with the γ-phosphate of GTP suggests a role in stabilizing the phosphoryl group undergoing transfer in a higher energy eclipsed conformation relative to the β-phosphoryl group, again making the γ-phosphate a better leaving group. The importance of Asn-403 in orienting Arg-405 to fulfill this role is also apparent from the structures (Fig. 2).

Unique nucleotide base-binding pocket found in mammalian PEPCK. Those residues discussed in the text are illustrated and labeled.

In the lid open PEPCK·GTP (4) and PEPCK·OAA·GTP (Fig. 2A) complexes, the bound conformation of GTP shows the orientation of the nucleotide ribose ring and α-phosphate in a non-canonical conformation. The structural studies suggest that this uncommon nucleotide orientation in the lid open state holds the nucleotide away from the M1 manganese ion, perhaps minimizing GTPase activity in the absence of a phosphoryl acceptor.

Activation of Catalysis upon Lid Closure

Although all of the substrates bind to the lid open form of the enzyme (Fig. 2A), in a clear example of a classical induced-fit mechanism, structural studies demonstrate that only upon lid closure are the substrates positioned correctly to allow for catalysis to proceed (5). This begs the question as to what is the energetic driving force for lid closure. The structural data support a model in which modulation of the free energy profile for the enzyme occurs as ligands bind during progression of PEPCK toward formation of the Michaelis complex (5). This model suggests that the thermodynamic favorability of the enzyme adapting the lid closed (active) state increases as ligands add to the enzyme, with a portion of their Gibbs free energy of binding being partitioned to the protein, offsetting the entropic unfavorability of the lid assuming an ordered closed conformation rather than a conformationally dynamic open (inactive) state. Therefore, after formation of the lid open Michaelis complex, the next step in the catalyzed reaction is the sampling of the lid closed state due to a thermodynamic shift in the favorability of its formation. Upon lid closure after formation of the Michaelis complex, the ribose ring and phosphate chain of GTP assume the canonical conformation stabilized by a shift in interaction between Arg-436 and the ribose ring from the C-3 hydroxyl group to the lactam oxygen (Fig. 2, compare A and B). This change in Arg-436/ribose ring interactions results in movement of the nucleotide forward in the active site toward the M1 ion, shortening the potential phosphoryl transfer distance by ∼0.5 Å. This shift in nucleotide position also results in a shift in the position of the M2 metal ion such that the M1-M2 metal distance decreases by ∼0.5 Å between the identically ligated lid open and lid closed complexes. Movement of the nucleotide and its associated metal ion appears to result from the shift in the position of the P-loop that is correlated with the transition of the lid from an open to a closed state. Interactions between the backbone amides of the P-loop and the β-phosphate of GTP move the nucleotide toward OAA and the M1 metal as the P-loop shifts toward the M1 metal. Based upon the structural studies, closure of the active-site lid, movement of the P-loop toward the M1 metal ion (closed P-loop conformation), and transition of the nucleotide from the unusual conformation observed in the lid open complexes to the canonical conformation observed in the lid closed state represent three interdependent processes. Formation of the lid closed conformation is associated with release of two potential steric constraints: 1) adaption by the P-loop of the closed conformation, which appears to be mediated via a hydrogen bond being established between Ser-286 and the C-4 oxygen of OAA, and 2) a change in the rotomeric conformation of Arg-436 upon nucleotide association (Fig. 2, compare A and B). Experimental evidence clearly demonstrates a correlation between the motions of the P-loop and the active-site lid, with both being essential for catalysis (35). Unfortunately, current data are insufficient to analyze these interdependent motions temporally; thus, a mechanism of cause and effect cannot yet be established. Nevertheless, the necessity for dynamic behavior of the P-loop indicated by the recent studies does provide a possible explanation for the inactivation of PEPCKs from many sources by chemical modification of Cys-288, which resides on the P-loop (14, 15).

The modeled position of CO2 in Fig. 2 (C and D), which is based upon the positions of the carboxylate of OAA and the sulfate of 3-sulfopyruvate (4, 5), mirrors the observed position of CO2 bound in the E. coli ATP-dependent form of PEPCK (23). Based upon the structural information, decarboxylation of OAA to form the enolate intermediate is facilitated by polarization of the carboxylate between Asn-403 and Arg-87 (Fig. 2C). In addition, the rotation forward of Tyr-235 provides further interactions with CO2, as does aromatic stabilization by the ring of Phe-333, neither of which is available in the complexes with OAA (Fig. 2, compare B and C). As described previously, phosphorylation of the resultant enolate occurs by positioning and stabilization of the phosphoryl group undergoing transfer through electrostatic interactions with Arg-405, Lys-290, and the two metal ions (Fig. 2C).

Lid Opening and Product Release

In the reverse of the process described above, the formation of products again results in modulation of the free energy profile for the enzyme such that the chemical transformation of the enolate intermediate to PEP results in a decrease in the thermodynamic favorability of the lid closed state, allowing the enzyme once again to sample the lid open conformation. Upon sampling of the lid open state, the PEP product shifts away from direct coordination to the M1 and M2 metal ions (Fig. 2, D–E). For this transition to occur, Tyr-235 must shift to its rearward orientation, and in the process, the changing rotomeric states of Tyr-235 and Phe-333 lessen the enzyme/CO2 interactions and allow for the release of the CO2 product. The loss of direct M1 coordination by PEP, which is consistent with NMR studies suggesting outer-sphere coordination of PEP in the PEPCK·Mn2+·PEP complex (24), results in water filling the three open coordination sites to the active-site metal and the one open site on the nucleotide metal (Fig. 2E). The interactions with Ser-286 on the P-loop are lost as PEP moves to an outer-sphere coordination geometry, allowing for an opening of the P-loop and the resultant shift of the nucleotide and M2 metals away from the M1 metal as Ser-286 orients toward solvent (Fig. 2E). The interactions between GDP and Lys-290 on the P-loop are minimized, with Lys-290 populating two conformations in the PEPCK·PEP·GDP open complex (Fig. 2E) as suggested by the analogous crystal structure with the PEP analog 2-phosphoglycolic acid (5). The shift of PEP to the outer-sphere complex, which has been observed in both human (2) and rat7 cytosolic and chicken mitochondrial (3) PEPCKs, is facilitated through a number of interactions unique to this conformation, which are depicted in Fig. 2E. The shift to outer-sphere coordination and P-loop and lid opening allow for PEP and GDP release and the subsequent transition back to the beginning of the catalytic cycle.

Stabilization of the Enolate Intermediate

All data, both structural and biochemical, are consistent with the PEPCK-catalyzed reaction proceeding through an enolate intermediate. Because of the enolate's reactivity, this chemical mechanism requires PEPCK to protect this intermediate from alternative chemistries, especially its protonation that results in formation of pyruvate. Structural data indicate that the lid closed state becomes more highly populated upon transition of the enzyme from the Michaelis complex to the intermediate complex (5), which is consistent with a role for the active-site lid in protecting the enolate intermediate from protonation and subsequent tautomerization to pyruvate. Through the lid closed conformation of the enzyme being the most stable conformational state for the protein upon formation of the intermediate complex (Fig. 2C), the dynamics of the lid element become an essential component of the reaction pathway. The extrapolation of these results suggests the intriguing possibility that by carrying out the reaction in a stepwise fashion, the individual chemical steps of this pathway are more favorable and have lower transition state barriers than other mechanistic possibilities. Intrinsic to this hypothesis is the requirement that the enzyme stabilizes the enolate intermediate from alternative chemistries that make this chemical mechanism impossible for the uncatalyzed reaction in solution.

Structural Explanations for Nucleotide Specificity

The reasons that mammalian PEPCK utilizes GTP or ITP, but not ATP, as a substrate are suggested by crystal structures of PEPCK·nucleotide complexes (Fig. 3) (25). Structurally, adenosine and guanosine nucleotides differ at the 2- and 6-positions of the purine ring. In combination with binding studies of inosine (lacking the C-2 amino group) and guanosine nucleotides, structural data show that the C-2 amino group of GTP does play a role in nucleotide binding, likely mediated through its interaction with Phe-525 (Fig. 3) (2). However, the ability of the mammalian PEPCKs to utilize inosine and guanosine nucleotides with similar kinetic efficiency indicates that the absence of a C-2 amino group in ATP is not the basis for selectivity of PEPCK. In GTP and ITP, the C-6 carbonyl functions as a hydrogen bond acceptor, which is not possible for the C-6 amino group of ATP. Consistent with this distinguishing feature, structures of PEPCK complexes show that the C-6 carbonyl of GTP forms hydrogen bonds with the backbone amide of Phe-530 and the side chain amide nitrogen of Asn-533 (Fig. 3). Furthermore, the amide of Asn-533 is held in position by hydrogen bonds with the indole nitrogen of Trp-527 and backbone carbonyl of Phe-530 (Fig. 3). These interactions prevent Asn-533 from assuming an alternative rotomeric state that could allow it to position its amide group to accept a hydrogen bond from the C-6 amino group of ATP. The other distinguishing feature of guanosine and inosine nucleotides is the ability of the purine ring to tautomerize. Although many studies have demonstrated that in solution these bases populate the keto form, the possibility exists that the physicochemical characteristics of the nucleotide-binding pocket of the mammalian PEPCKs, which sandwiches the purine ring between Phe-333 and Phe-517 (Fig. 3), may preferentially impose selectivity for the enol tautomer unique to inosine and guanosine nucleotide bases. Further studies will be necessary to firmly establish in detail structural bases for different nucleotide specificities of the two general classes of PEPCK isozymes.


Taken as a whole, structural data indicate that the PEPCK active site creates an electrostatic environment tailored to stabilize the large amount of localized negative charge that results from close juxtaposition of its multiple anionic substrates. This environment facilitates stabilization of the two transition states as well as the enolate intermediate and allows for efficient catalysis. In concert with this electrostatic environment, unique dynamic properties associated with the reaction pathway allow for the enzyme to minimize off-pathway reactions, such as nucleotide hydrolysis and OAA decarboxylation, that would decrease catalytic fidelity.

Supplementary Material

Author Profile:

*This is the fourth article of four in the Thematic Minireview Series on the Biology of Phosphoenolpyruvate Carboxykinase: 55th Anniversary. This minireview will be reprinted in the 2009 Minireview Compendium, which will be available in January, 2010.

3Both ATP- and GTP-dependent PEPCK enzymes exist in nature. The ATP class has been identified in bacteria, trypanosomatids, C4 plants, and yeast, whereas the GTP-dependent isozymes are found in mammals and other eukaryotes (excluding yeast) and some bacteria (Corynebacterium sp. and Mycobacterium sp.) (8, 25, 26).

4In the absence of substrates, PEPCK binds a single divalent cation cofactor (M1). Manganese is the most activating cation and is thought to be the physiologically relevant cofactor due to this property in combination with the higher affinity of PEPCK for manganese relative to magnesium. An additional divalent cation (M2) is bound in the presence of nucleotide, as the catalytically competent form of the nucleotide is the M2+·nucleotide complex.

5The atom numbering utilized is consistent with that used previously (4). Although this is contrary to IUPAC nomenclature, it is consistent with the atom numbering utilized for the molecules in the Protein Data Bank.

6The residue numbering utilized is that of the rat cytosolic isozyme.

7T. Holyoak, unpublished data.

2The abbreviations used are:

phosphoenolpyruvate carboxykinase


1. Yang J., Kalhan S. C., Hanson R. W. (2009) J. Biol. Chem. 284, 27025–27029 [PMC free article] [PubMed]
2. Dunten P., Belunis C., Crowther R., Hollfelder K., Kammlott U., Levin W., Michel H., Ramsey G. B., Swain A., Weber D., Wertheimer S. J. (2002) J. Mol. Biol. 316, 257–264 [PubMed]
3. Holyoak T., Sullivan S. M., Nowak T. (2006) Biochemistry 45, 8254–8263 [PubMed]
4. Sullivan S. M., Holyoak T. (2007) Biochemistry 46, 10078–10088 [PubMed]
5. Sullivan S. M., Holyoak T. (2008) Proc. Natl. Acad. Sci. U.S.A. 105, 13829–13834 [PMC free article] [PubMed]
6. Trapani S., Linss J., Goldenberg S., Fischer H., Craievich A. F., Oliva G. (2001) J. Mol. Biol. 313, 1059–1072 [PubMed]
7. Cotelesage J. J., Prasad L., Zeikus J. G., Laivenieks M., Delbaere L. T. (2005) Int. J. Biochem. Cell Biol. 37, 1829–1837 [PubMed]
8. Aich S., Prasad L., Delbaere L. T. (2008) Int. J. Biochem. Cell Biol. 40, 1597–1603 [PubMed]
9. Matte A., Goldie H., Sweet R. M., Delbaere L. T. (1996) J. Mol. Biol. 256, 126–143 [PubMed]
10. Matte A., Tari L. W., Delbaere L. T. (1998) Structure 6, 413–419 [PubMed]
11. Sudom A. M., Prasad L., Goldie H., Delbaere L. T. J. (2001) J. Mol. Biol. 314, 83–92 [PubMed]
12. Tari L. W., Matte A., Goldie H., Delbaere L. T. J. (1997) Nat. Struct. Biol. 4, 990–994 [PubMed]
13. Tari L. W., Matte A., Pugazhenthi U., Goldie H., Delbaere L. T. (1996) Nat. Struct. Biol. 3, 355–363 [PubMed]
14. Lewis C. T., Seyer J. M., Carlson G. M. (1989) J. Biol. Chem. 264, 27–33 [PubMed]
15. Makinen A. L., Nowak T. (1989) J. Biol. Chem. 264, 12148–12157 [PubMed]
16. Stiffin R. M., Sullivan S. M., Carlson G. M., Holyoak T. (2008) Biochemistry 47, 2099–2109 [PubMed]
17. Ash D. E., Emig F. A., Chowdhury S. A., Satoh Y., Schramm V. L. (1990) J. Biol. Chem. 265, 7377–7384 [PubMed]
18. Guidinger P. F., Nowak T. (1990) Arch. Biochem. Biophys. 278, 131–141 [PubMed]
19. Castillo D., Sepúlveda C., Cardemil E., Jabalquinto A. M. (2009) Biochimie 91, 295–299 [PubMed]
20. Konopka J. M., Lardy H. A., Frey P. A. (1986) Biochemistry 25, 5571–5575 [PubMed]
21. Sheu K. F., Ho H. T., Nolan L. D., Markovitz P., Richard J. P., Utter M. F., Frey P. A. (1984) Biochemistry 23, 1779–1783 [PubMed]
22. Krautwurst H., Bazaes S., González F. D., Jabalquinto A. M., Frey P. A., Cardemil E. (1998) Biochemistry 37, 6295–6302 [PubMed]
23. Cotelesage J. J., Puttick J., Goldie H., Rajabi B., Novakovski B., Delbaere L. T. (2007) Int. J. Biochem. Cell Biol. 39, 1204–1210 [PubMed]
24. Hebda C. A., Nowak T. (1982) J. Biol. Chem. 257, 5515–5522 [PubMed]
25. Hanson R. W., Patel Y. M. (1994) Adv. Enzymol. Relat. Areas Mol. Biol. 69, 203–281 [PubMed]
26. Mukhopadhyay B., Concar E. M., Wolfe R. S. (2001) J. Biol. Chem. 276, 16137–16145 [PubMed]

Articles from The Journal of Biological Chemistry are provided here courtesy of American Society for Biochemistry and Molecular Biology

Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...