Characterization of 3‑[(Carboxymethyl)thio]picolinic Acid: A Novel Inhibitor of Phosphoenolpyruvate Carboxykinase
Phosphoenolpyruvate carboxykinase (PEPCK) is a meta- bolic enzyme catalyzing the reversible interconversion of oxaloacetic acid (OAA) and phosphoenolpyruvate (PEP) (Scheme 1). Two classes of the enzymes exist depending on whether they use ATP/ADP or GTP/GDP as the phosphoryl donor/acceptor pair. PEPCK has also been demonstrated to have a requirement for two divalent metal cations.1−3 In the GTP-dependent enzymes being studied here, Mn2+ (M1) is typically observed to be the most activating cation and acts as a true metal cofactor and coordinates with OAA in the active site. A second divalent cation (M2), typically Mg2+, is also required to stabilize the γ- and β-phosphates of the bound nucleotide triphosphate, and as such, the Mg2+-nucleotide acts as the true substrate. Structural studies have demonstrated that the transferable phosphoryl group in this freely reversible reaction acts as a bridging ligand between the two metal ions.4,5 The coupling of the catalyzed reaction to conforma- tional changes in PEPCK has been elucidated through various structural studies.4,5 Once the Michaelis complex is formed, the enzyme prefers a closed conformation. This catalytically active state is coincident with OAA decarboxylation, resulting in the formation of a reactive enolate intermediate that is subsequently phosphorylated via an in-line phosphoryl transfer by the nucleoside triphosphate, a result that is consistent with the existing body of kinetic evidence6,7(Scheme 1).
From the perspective of its biological function, PEPCK has traditionally been classified as a gluconeogenic enzyme whose role in metabolism is to remove OAA from the citric acid cycle for the purpose of de novo glucose synthesis (reviewed in refs 8−10). More recently, studies have shown that PEPCK’s metabolic importance might be underappreciated, as its role as a general cataplerotic enzyme may result in its function being as a master regulator of TCA cycle flux. Support for such a role for PEPCK comes from studies that demonstrate PEPCK activity levels can be linked to processes as diverse as cancer cell proliferation,11 initiation and sustainability of Mycobacte- rium tuberculosis infections,12 glucose-stimulated insulin secretion in diabetes,13 an increase in aerobic capacity,14 and cellular aging.15
Due to PEPCK’s postulated role in these diverse biological processes in various organisms, the regulation of its activity should be essential in maintaining homeostasis. Regulation of PEPCK activity has historically been assigned to mechanisms operating at the transcriptional level.16−19 Only recently have any mechanisms of regulation at the protein level been suggested. It has been proposed that PEPCK is subject to lysine acetylation that targets the enzyme for ubiquitination and subsequent degradation.20−22 In more recent work, the same acetylation, as well as phosphorylation, has been suggested to modulate the predominant direction of the reversible reaction in vivo.23
The reversible inhibition of PEPCK by various small molecules has been examined by many groups over the past 50 years. These studies have largely focused on either aImage rendered in ChemDraw Pro 11.0. analogues of PEP/OAA,24−26 tryptophan metabolites,27−32 or nucleotide analogues.33,34 Focusing upon the PEP/OAA binding site, in 2008 Stiffin et al. used various commercially available small molecules to understand the physicochemical properties required for inhibition of PEPCK.35,36 These small molecules were selected with the rationale that they were analogues of OAA and PEP. Structural and kinetic studies demonstrated that two inhibitors, sulfoacetate and oxalate, defined two partially overlapping binding pockets in the PEP/ OAA binding site; these two sites were defined as the inner and outer subsites. The inner subsite, as defined by the location of oxalate/OAA binding, is a result of direct coordination of the ligand with the M1 metal ion, which is coincident with the rotation of the phenolic side chain of Y235 in a direction toward the M1 cation. In contrast, the outer subsite occupied by sulfoacetate is generated when Y235 occupies a rotomeric conformation that rotates the phenolic side chain in a direction away from the M1 metal. This outer subsite was originally described by the binding of the substrate PEP in a second-sphere coordination geometry with the phosphate of PEP interacting with the M1 metal ion indirectly through intervening water molecules.37 Taken together, the prior studies raised the intriguing possibility of whether both binding pockets could be simultaneously occupied by a single inhibitor molecule. On the basis of these data, Stiffin et al. put forth design criteria that suggested a rational basis for the structure-based design of more selective PEPCK inhibitors focused upon the partially non-overlapping inner and outer subsites. In concert with this original work, a recent study from our laboratory provided the structural basis for the complex inhibition exhibited by the well-studied inhibitor of PEPCK, 3- mercaptopicolinic acid (MPA).38 Since its original characterization in the 1970s, MPA has been used as a potent inhibitor of PEPCK both in vivo and in vitro.28,31,32,38−41 The recent structural and kinetic data are consistent with a mechanism of inhibition in which MPA binds to two different sites on PEPCK. One site utilizes direct coordination of the pyridine nitrogen and the exocyclic carboxylate group of MPA with the M1 active site manganese ion and is therefore similar to the binding of OAA/oxalate at the inner subsite. A secondary binding site was also identified as it was observed that MPA bound to a site behind the P-loop, a necessary mobile element of active site architecture. Despite the previous widespread use of MPA as an in vivo and in vitro selective inhibitor of PEPCK function, the inhibitor possesses some less than optimal properties. First, the complex kinetics observed due to the binding of two molecules of MPA in non-overlapping binding sites (noncompetitive and competitive) can make the resultant kinetic data difficult to interpret.31,39 Second, the free sulfhydryl group of MPA, a property that has been demonstrated to be key to its potency, can result in the formation of intermolecular mixed disulfide bonds between the inhibitor and reactive protein cysteine side chains in the absence of a reducing environment.42 Supporting this idea are studies that demonstrate that MPA can inactivate PEPCK by forming a covalent mixed disulfide bond with a the well- characterized reactive cysteine (C288, rat cPEPCK number- ing) located on the P-loop of the enzyme.38 Lastly, while MPA has been suggested to exhibit promiscuity by reversibly inhibiting glucose 6-phosphatase (G6Pase), itself an essential enzyme in the de novo synthesis of glucose along the gluconeogenic pathway, the observation of a time dependence of the observed inhibition coupled with other in vivo studies may suggest that the activity loss in G6Pase is due to the formation of a mixed disulfide between
G6Pases and MPA, similar to what is observed with PEPCK.43−45
Utilizing the design criteria outlined in the study by Stiffin et al. and the structure−function data of MPA inhibition of cPEPCK, we have synthesized 3-[(carboxymethyl)thio]- picolinic acid (CMP) (Figure 1) using a novel chemical scheme (Scheme 2) as a potentially selective and potent inhibitor of PEPCK. The structural and kinetic data presented here demonstrate that as designed, CMP inhibits rat cytosolic (rcPEPCK), human mitochondrial (hmPEPCK), and M. tuberculosis (mtbPEPCK) enzymes via a competitive mecha- nism. In addition, the extended carboxymethyl “tail” of CMP results in simultaneous occupancy of both inner and outer subsites by a single molecule and provides support for the approach of designing selective inhibitors of PEPCK through targeting the extended binding pocket as originally postu- lated.35
MATERIALS AND METHODS
Materials. Nickel-NTA resin was purchased from UBPBio, while P6-DG and CHT ceramic hydroxyapatite type 2 resin were purchased from Bio-Rad. TCEP and DTT were purchased from Gold Biotechnology. PEP and NADH were purchased from Chem-Impex International. Malate dehydro- genase was purchased from Calzyme. GDP was purchased from Sigma, while HEPES was purchased from Fisher Scientific. SUMO protease was expressed and purified as previously described.46 All chemicals required for the synthesis of CMP were purchased from Sigma-Aldrich Canada and used
as received. All other materials were purchased at the highest grade available.
Expression and Purification. Enzyme Expression. The genes for all three enzymes were cloned into the pe_SUMOstar (Kan) vector (LifeSensors), resulting in the proteins of interest being expressed as His6-SUMO fusions.The cloning of rat cPEPCK has been described previously.46 The translated sequence of mtbPEPCK (Rv0211) was used to guide codon-optimized gene synthesis for Escherichia coli expression (Blue Heron Biotechnology), and this gene was subsequently cloned into the BsaI and XhoI sites of pe_SUMOstar (Kan). Upon cleavage from the N-terminal SUMO fusion, the final mtbPEPCK product encompasses the full native sequence of the enzyme with no additional residues. The human PCK2 gene was purchased from ATCC (GenBank entry BC001454) and cloned into the BsaI and XhoI sites of pe_SUMOstar (Kan). The cloned fragment of the human PCK2 gene lacks the DNA sequence encoding the mitochondrial localization sequence (amino acids M1−V32), and upon liberation from the SUMO fusion, the final protein product begins at L33, consistent with the native protein after mitochondrial import. All plasmid constructs were transformed into E. coli BL-21(DE3) cells for protein expression. All expressions were carried out at 20 °C for 16−24 h in ZYP- 5052 autoinduction medium.47 The cells were harvested by centrifugation at 5000g for 10 min followed by subsequent storage of the resultant cell pellets at −80 °C.
Fractions containing hmPEPCK were collected and buffer exchanged back into 5 mM KPO − and 2 mM TCEP-HCl before SUMO fusion proteolysis for 1 h with 2.5 mg of SUMO protease. The digestion mixture was subsequently incubated mL fractions with 25 mM HEPES (pH 7.5), 300 mM imidazole, and 2 mM TCEP-HCl. All fractions containing PEPCK were concentrated below 10 mL with a stirred cell Amicon concentrator and 30 kDa mass cutoff Millipore filter. PEPCK was then buffer exchanged into 25 mM HEPES (pH 7.5) and 2 mM TCEP-HCl via passage through a P6DG desalting column. Fractions containing PEPCK were incubated overnight with 2.5 mg of SUMO protease purified as previously described.46
After liberation of the SUMO fusion and 2 mM TCEP-HCl for 1 h. Significant precipitation was observed during this 1 h incubation period, and as a consequence, total hmPEPCK yields diminished significantly during this step. The nickel-NTA was washed with 5 mM KPO4− and 2 mM TCEP-HCl, and the flow-through containing hmPEPCK was collected and buffer exchanged into 5 mM KPO − (pH 6.8) and 10 mM DTT. This solution was concentrated to 2 mg/mL using an ε280 of 1.63 mL/mg (115390 M−1 cm−1) and flash-frozen in 20 μL aliquots by direct immersion in liquid nitrogen. These pellets were subsequently stored at −80 °C.
Structure Determination. The structure of rcPEPCK in complex with CMP was determined using the molecular replacement method using the rcPEPCK holoenzyme (PDB entry 2QEW) as a molecular replacement model in MOLREP from the CCP4 package.50 Manual adjustments and ligand, water, and metal fitting were carried out using COOT.51,52 Refinement of the model was carried out with REFMAC5 from the CCP4 package.53 Final model validation was completed using Molprobity (http://molprobity.biochem.duke.edu).54,55
■ RESULTS AND DISCUSSION
As described above, the 2008 Stiffin investigation presented structure-based design criteria that were postulated to provide a framework for the design of potentially potent and selective inhibitors of GTP-dependent PEPCKs.35 In brief, that study concluded that anionic bifunctional acids were most potent and that these molecules could either directly coordinate in the inner subsite with the M1-Mn2+ metal cofactor as seen with oxalate, picolinic acid derivatives, and OAA or bind to the outer sphere PEP binding subsite. Important for the current work, designing more potent or selective inhibitors was hypothesized to be possible if a single molecule could be synthesized that bridged both of the inner and outer subsites, a feature that has not been exhibited to date by a single molecular scaffold. To investigate whether both subsites could indeed be occupied simultaneously, we set out to design a molecule that followed the guidelines presented in that prior work. On the basis of the recent structural characterization of the potent PEPCK inhibitor MPA, we chose to use the picolinic acid scaffold as the starting point for the design of such a molecule, resulting in the proposal to synthesize CMP.38
While the synthesis of CMP has been previously proposed, difficulties encountered in the synthesis of CMP by the reported literature method involving diazotization of 3- aminopicolinic acid followed by treatment with thioglycolic acid57 resulted in the pursuit of an alternative method. Initially, the preparation of CMP was thought to occur through the base-induced addition of methyl mercaptoacetate to 3- bromopyridine-2-carbonitrile, followed by hydrolysis of the nitrile and ester groups. However, attempts to perform this reaction with NaH or DBU gave none of the intended product, furnishing instead the product of intramolecular cyclization and tautomerization of the initial adduct (Scheme 2A). Subsequently, it was found that the addition of the carboxymethylmercapto moiety could be affected without further reaction by using thioglycolic acid in place of the ester and treating it with 2 equiv of DBU in THF followed by acidic aqueous workup. Finally, hydrolysis of the intermediate adduct nitrile was effected in refluxing 6 M HCl (Scheme 2B). This protocol was found to be successful in generating the desired CMP and verified by crystallization and structure solution by X-ray crystallography (Figure 2) as well as spectroscopic analysis (Figures S1−S5).
Synthesis of CMP. Structural Characterization of the rcPEPCK−CMP Complex. After a reliable synthetic scheme was developed, structural studies of the compound in complex with rcPEPCK were initiated to ascertain whether the compound bound to the active site of PEPCK as hypothesized. As all GTP-dependent enzymes active sites are highly conserved, the binding of CMP to rcPEPCK is hypothesized to be representative of its binding to all GTP-dependent PEPCKs. The co-crystal structure of rcPEPCK in complex with CMP was determined to 1.5 Å resolution, and the final data and model statistics for the structure are listed in Table 1. The complex between rcPEPCK and CMP was found to crystallize in the P21 space group and contains one molecule of PEPCK in the asymmetric unit (ASU). Similar to the complex between rcPEPCK and either MPA, OAA/PEP, or OAA/PEP analogues, the enzyme is observed to populate the “open” conformation as typified by both a disordered Ω-loop and the observation that the N- and C-terminal domains are rotated away from one another effectively increasing the size of the active site cleft.35,37,48 Clear electron density for CMP is present in the OAA/PEP binding site allowing for unambiguous modeling of the ligand in the binding pocket (Figure 3B). Similar to the rcPEPCK−PEP complex, and in contrast to the BSP/OAA/oxalate complexes,46,48 Y235 is observed to rotate away from the M1-Mn2+ ion, increasing the size of the OAA/PEP binding pocket and allowing the CMP molecule to bridge both inner (OAA) and outer (PEP) subsites with the carboxylate of the CMP molecule making the same edge-on interaction with Y235 as the carboxylate of PEP (Figure 3A).
A comparative analysis of the PEPCK−CMP complex presented here with the prior structures of rcPEPCK in complex with either sulfoacetate or MPA demonstrates that CMP binds to the active site pocket of PEPCK exactly as predicted from the superpositioning of the independent sulfoacetate and MPA complexes (Figures 3 and 4). Importantly, the rcPEPCK−CMP complex demonstrates for the first time that occupancy of the inner and outer subsites simultaneously by a single molecule is possible. Unsurprisingly, CMP’s picolinic acid structure is found to be coordinated to the active site manganese cation at the inner site by one oxygen of the exocyclic 2-carboxylate and the pyridine nitrogen of the ring moiety in a fashion identical to that of the prior MPA complex (Figures 3B and 4A). Doing so fulfills the first criterion highlighted in the Stiffin investigation of a cis-planar sp2 hybridization.35 Beyond the expected metal coordination through the picolinic acid ring, in general, CMP makes extensive interactions with key, conserved active site residues that are similar in nature to the individual contacts exhibited by the sulfoacetate and MPA ligands in those complexes (Figure 4). As mentioned previously, MPA’s sulfhydryl moiety acts a bridge between sulfoacetate and MPA to create CMP. This thioether satisfies the third design criterion of Stiffin et al. for an electron rich atom at the C3 position of OAA and makes interactions with both R87 and the guanidinium nitrogen of R405. Unfortunately, the neutral sulfur of the CMP thioether does not have the ability to exploit electrostatic interaction with R87 and R405 as seen with the binding of SA. Attempts to oxidize the sulfur to potentially utilize these interactions led to a molecule that failed to inhibit rcPEPCK at concentrations of <1 mM. This failure is likely due to the steric conflict between the sulfate oxygens and the requirement for a planar geometry of the exocyclic carboxylate of the picoloinc acid (unpublished results). Looking toward the outer sphere PEP binding site, we found previous structural data have demonstrated that this site is utilized for both PEP binding prior to closure of the active site lid as well as sulfoacetate binding.4,36 In addition, the body of data also suggests that gating of this secondary binding pocket is regulated by the rotomeric state of Y235. Previously, ligands as typified by oxalate that occupy the inner subsite are stabilized by an edge-on interaction with the phenolic ring of the forward rotomer of Y235, which in turn closes off the outer subsite. For the outer subsite to be opened to the inner subsite, the phenol ring of Y235 must undergo a 4.4 Å tip-to-tip conformational shift away from the M1 metal as is seen in the complex with CMP (Figure 3A). The predominant conserved interactions in the outer subsite at the outermost limits of the subsite are between the carboxylate moiety of either sulfoacetate or PEP and the side-chain amide of N403 and the backbone amides of G237 and R87, which the new structural data confirm are conserved in the rcPEPCK−CMP complex (Figure 4). Kinetic Characterization of the Inhibition of PEPCK by CMP. Because the structural data indicated that CMP was able to simultaneously bridge the inner and outer sphere OAA/PEP binding subsites of rcPEPCK and by doing so fulfilled the major structure-based design criteria set out by Stiffin et al., we next undertook kinetic investigations to assess the potency of the binding of the molecule to PEPCK. In addition, to assess the ability of CMP to act as a general inhibitor of PEPCK, three isozymes were tested for inhibition by CMP: the rat cytosolic enzyme (rcPEPCK), the human mitochondrial enzyme (hmPEPCK), and the enzyme from M. tuberculosis (mtbPEPCK). Initially, IC50 values were determined for CMP against both rcPEPCK and hmPEPCK at a fixed PEP concentration of 500 μM to determine if consistent with the structural data, the compound was able to inhibit PEPCK at a reasonable concentration prior to carrying out full characterization of the inhibition. While CMP was able to inhibit both forms of PEPCK with IC50 values in the range of ∼100 μM, this potency was approximately 1 order of magnitude lower than that observed for MPA (Table 2 and Figure S6). Consistent with the IC50 values, the calculated Ki values from these data are 29 and 53 μM for rat cytosolic and human mitochondrial PEPCK, respectively. For comparison, the Ki for MPA against hmPEPCK was found to be ∼3 μM (Table 2), while the Ki against the rat enzyme has previously been determined to be 8 μM.38 Complete kinetic characterization of the inhibition by CMP of the human, tuberculosis, and rat enzymes was conducted in the OAA-forming direction with PEP as the variable substrate (Table 3 and Figure S7). These inhibition constants were calculated from replots of the kinetic parameters determined from complete kinetic profiles at fixed, variable concentrations of CMP. This approach, rather than utilizing a global fit of all of the data, was utilized as replots of 1/Vmaxapp versus [I] showed a slight intercept effect (data not shown) that is an outcome of picolinic acid derivatives having a general ability to chelate divalent metal cations that are required for the formation of the catalytically relevant metal−nucleotide substrate and PEPCK-M1 holoenzyme.58 Decreasing the level of M1 metal leads to an apparent decrease in the active enzyme concentration and thereby presents as noncompetitive inhibition. The Ki values determined from the full kinetic profiles were found to be 29, 55, and 35 μM for rat cytosolic, M. tuberculosis, and human mitochondrial PEPCK, respectively (Table 3). As expected for a competitive mechanism of inhibition, these Ki values correspond very well with those values determined from the IC50 experiments at a fixed, single concentration of PEP and again suggest that the potency of CMP is in the range of 30−50 μM that corresponds to a potency of inhibition that is approximately 1 order of magnitude lower than that of MPA. Comparing the difference in CMP’s potency against the three tested isozymes shows that as one would expect on the basis of the conservation of the contacting residues among the three enzymes there is a minimal although measurable difference in the potency of the compound among the three enzyme forms. On the basis of the knowledge that all of the contacting residues are conserved and extensive insight into the coupling of conformational changes in the enzymes to the binding of ligands at the active site, these small differences may simply reflect the isozyme specific differences in the energetic costs of the disorder to order transitions that are coupled to substrate/inhibitor binding due to sequence and structural differences away from the immediate binding surfaces. It is also interesting to note that on the basis of the prior characterization of Ki values for the inhibition of rcPEPCK by MPA and sulfoacetate in the OAA forming direction the Ki observed for CMP inhibition of rcPEPCK is, to a first approximation, largely an average of the two individual Ki values of the building blocks of MPA and sulfoacetate (CMP Ki = 29 ± 3 μM, sulfoacetate Ki = 83 ± 5 μM,35 and MPA Ki = 8 ± 1 μM38). As mentioned above, this result is not entirely surprising, as CMP lacks the sulfate group of SA and the corresponding electrostatic interactions with R87 and R405. Importantly, these new data demonstrate that despite the intricate coupling of conformational change and disorder to order transitions that are linked to ligand binding by PEPCK, simultaneous occupancy of the inner and outer PEP/OAA binding pockets is possible with a single molecular scaffold. In addition to supporting the notion that the inner and outer subsites of the PEP/OAA binding pocket can be occupied simultaneously, the presence of the CMP thioether linkage over the free reactive sulfhydryl group of MPA will prevent the formation of mixed disulfide linkages with free cysteine residues on PEPCK that has been demonstrated to lead to covalent inactivation of the enzyme in vitro.38,59−61 Additionally, if the presence of the free sulfhydryl on MPA is the source of the observed time-dependent inhibition of G6Pase, this modification could support the utility of CMP or CMP-like molecules over MPA in future in vivo metabolic studies. In conclusion, we present here the synthesis and character- ization of a novel inhibitor of PEPCK that spans the inner and outer subsites of the PEPCK active site simultaneously. While the molecule does not exhibit an improved potency over MPA, the scaffold represents a reasonable starting point for additional structure-based inhibitor design. One avenue for future modifications could include bridging CMP toward the nucleotide binding site to couple with previously determined nucleotide analogues,33,34 effectively occupying the entirety of the active site cleft and utilizing SKF-34288 the binding energy associated with both active site divalent cations.