An allosteric site at the biotin carboxylase dimer interface mediates activation and inhibition in Staphylococcus aureus pyruvate carboxylase

Abstract

Allosteric regulation of the essential anaplerotic enzyme, pyruvate carboxylase (PC), is vital for metabolic homeostasis. PC catalyzes the bicarbonate- and ATP-dependent carboxylation of pyruvate to form oxaloacetate. Dysregulation of PC activity can impact glucose and redox metabolism which contributes to the pathogenicity of many diseases. To maintain homeostasis, PC is allosterically activated by acetyl-CoA and allosterically inhibited by l-aspartate. In this study, we further characterize the molecular basis of allosteric regulation in Staphylococcus aureus PC (SaPC) using slowly/non-hydrolysable dethia analogs of acetyl-CoA and site-directed mutagenesis of residues at the biotin carboxylase homodimer interface. The dethia analogs fully activate SaPC but demonstrate significantly reduced binding affinities relative to acetyl-CoA. Residues Arg₂₁, Lys₄₆ and Glu₄₁₈ of SaPC are located at the biotin carboxylase dimer interface and play a critical role in both allosteric activation and inhibition. A structure of R21A SaPC in complex with acetyl-CoA reveals an intact molecule of acetyl-CoA bound at the allosteric site, offering new molecular insights into the acetyl-CoA binding site. This study demonstrates that the biotin carboxylase domain dimer interface is a critical allosteric site in PC, serving as a convergence point for allosteric activation by acetyl-CoA and inhibition by l-aspartate.

Graphical Abstract

INTRODUCTION

Pyruvate carboxylase (PC; EC 6.4.1.1) catalyzes the biotin- and ATP-dependent carboxylation of pyruvate to oxaloacetate. This reaction serves as an important step in replenishing citric acid cycle intermediates and supporting anabolic pathways. Dysfunctions in PC have been associated with conditions such as cancer, type II diabetes, and bacterial infection (1-4). Notably, in Staphylococcus aureus, PC has been found to contribute to virulence in both systemic and abscess infections (5).

Staphylococcus aureus PC (SaPC), similar to other PC enzymes, is a homotetramer composed of four distinct domains within each subunit: the biotin carboxylase (BC) domain, the carboxyl transferase (CT) domain, the biotin carboxyl carrier protein (BCCP) domain, and the central allosteric domain (AD) (Figure 1). The biotin cofactor, covalently tethered to the BCCP domain, undergoes MgATP-dependent carboxylation in the BC domain with HCO₃⁻ serving as the carboxyl group donor. The tethered carboxybiotin cofactor then transitions to the CT domain, where the carboxyl group is liberated as CO₂ and is attacked by the pyruvate enolate to generate oxaloacetate. Inter-subunit carrier domain movement and communication plays a crucial role in catalysis by facilitating intermediate channeling between distant active sites. Ongoing research is focused on unraveling the specific mechanisms responsible for inter-domain coordination and the role of allosteric effectors in mediating this process (6-9).

Figure 1. SaPC reaction and structure.

The half reactions for the BC domain and CT domain are shown with reference to the active site locations. On the left, the tetramer is illustrated with the two top subunits in cartoon representation (PDB 3BG5). One of the two subunits is outlined in surface representation. The bottom face of the tetramer is illustrated in grey surface representation. The allosteric binding site, with a bound molecule of CoA illustrated as spheres, is circled (superimposed from PDB 3HO8). BC is colored blue, CT is colored yellow, AD is colored green, and BCCP is colored red. On the right, the tetramer is rotated 90° to show the BC dimer interface, where the two BC domains (one from each subunit) are illustrated in different shades of blue.

PC, like many central metabolic enzymes, is subject to extensive allosteric regulation. Acetyl-CoA serves as a non-essential allosteric activator of PC in nearly all organisms, while l-aspartate and l-glutamate act as allosteric inhibitors in microorganisms and vertebrates, respectively (10). The degree of sensitivity to allosteric activation by acetyl-CoA depends on the organism, with the $k_{\text{cat}}$ increasing up to 35-fold in the presence of acetyl-CoA (11). In SaPC, the $k_{\text{cat}}$ increases ~5-10 fold in the presence of acetyl-CoA, and decreases ~2-5 fold in the presence of l-aspartate (9,12). Acetyl-CoA binds near the intersection of the AD and the BC domain dimer interface (Figure 1) and multiple interactions with the phosphates in the 3'5'-ADP portion of the activator contribute to acetyl-CoA binding (13). In addition, the acetyl moiety makes an important contribution to enzyme activation. While both CoA and acetyl-CoA can fully activate the enzyme, acetyl-CoA exhibits a ~100-fold greater affinity than CoA in SaPC (12). The molecular basis for the enhanced affinity afforded by the acetyl moiety is unclear. Unfortunately, in all X-ray crystal structures of PC determined with acetyl-CoA, the electron density has not revealed the presence of the acetyl moiety, precluding a complete description of the molecular interactions contributing to allosteric activation. It can be inferred from the structures that the acetyl moiety binds at the dimer interface of the biotin carboxylase (BC) domain, implicating the BC domain interface as a major contributor to allosteric activation (14). Consistent with this binding locus, acetyl-CoA enhances the binding affinity of substrates and metal ions to the BC domain, resulting in increased ATP cleavage efficiency and improved coupling of the BC and CT domain half-reactions (15).

Similar to acetyl-CoA, the primary locus of allosteric inhibition by l-aspartate is the BC domain. l-Aspartate inhibits bicarbonate-dependent ATP cleavage in the BC domain, while leaving the CT domain reaction unaffected (16). The inhibition mechanism is thought to involve either the inhibition of carboxybiotin formation or the disruption of carrier domain translocation (17). Interestingly, l-aspartate is a competitive inhibitor with respect to acetyl-CoA, suggesting that both allosteric effectors share an overlapping binding site (12,16,17). However, mutagenesis studies targeting residues near the binding site for the nucleotidyl portion of acetyl-CoA demonstrated a reduction in acetyl-CoA activation without affecting the degree of l-aspartate inhibition (13,17). While this suggests that the two effector molecules may have distinct binding sites, the potential for direct overlap between l-aspartate and acetyl-CoA at the acetyl moiety binding site has not been directly investigated.

Here, we propose that the acetyl moiety of acetyl-CoA is positioned at the BC dimer interface. We further propose that this binding site represents the site of overlap between l-aspartate and acetyl-CoA. Using acetyl-CoA analogs, site-directed mutagenesis and X-ray crystallography, we probe the structure and function of the acetyl moiety binding site at the BC dimer interface and demonstrate an important contribution from the acetyl thioester of the allosteric activator, along with three critical residues (Lys₄₆, Glu₄₁₈, and Arg₂₁) that lie in close proximity to the acetyl moiety. We demonstrate that these residues mediate allosteric activation and that they are required to enable allosteric inhibition by l-aspartate. A single mutation at the BC dimer interface, E418A, was sufficient to eliminate both allosteric activation by acetyl-CoA and allosteric inhibition by l-aspartate. These findings expand the definition of allostery in PC to include an important allosteric site at the BC dimer interface.

MATERIALS and METHODS

Materials.

Acetyl-CoA was purchased from CoALA Biosciences (Austin, TX). Except where noted, all other chemicals and materials were obtained from Sigma-Aldrich Chemical Co. (St. Louis, MO), Fisher Scientific Inc. (Pittsburgh, PA), or Research Products International Corp. (Mount Prospect, IL). All DNA oligonucleotide primers were obtained from Integrated DNA Technologies (Coralville, IA). Pyruvate carboxylase from S. aureus3 (SaPC; UniProtKB - A0A0H3JRU9) was previously cloned into a modified pET-27b vector and was generously supplied by Dr. Liang Tong, Columbia University (12). To reclone SaPC into the pTXB1 vector (New England Biolabs, Ipswich, MA), the gene encoding SaPC in pET-27b was PCR amplified using forward (5'- GCC ATA TGA AAC AAA TAA AAA AG -3') and reverse primers (5'-CGT GAT GCA GTT AGT TGC TTT TTC AAT TTC G -3'). The PCR amplicon was subjected to restriction digestion with NdeI and SapI prior to ligation into a gel-purified pTXB1 vector digested with NdeI and SapI. WT SaPC pTXB1 was fully sequenced and was found to include a PCR-introduced G571S mutation. The mutation was corrected back to a glycine at residue 571 by mutagenesis using the whole-plasmid PCR technique according to the Quikchange II mutagenesis protocol from Agilent Technologies, Inc. (Santa Clara, CA), with forward primer 5’- GCG GAC GTA TTT AAA GAT GGT TTC TCA CTA G -3’ and reverse primer 5’- CTA GTG AGA AAC CAT CTT TAA ATA CGT CCG C -3’. The complete gene sequence of WT SaPC in the pTXB1 vector was confirmed by DNA sequencing.

Synthesis of acetyl-CoA analogs.

The analogs acetyl-oxa(dethia)-CoA and acetyl-aza(dethia)-CoA were synthesized using a previously published chemoenzymatic approach (18). A similar chemoenzymatic approach was taken to generate acetyl-carba(dethia)-CoA. Briefly, 5-amino-2-pentanone ethylene ketal was condensed with previously prepared pantothenic acid acetonide in dichloromethane with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide to yield ethylene ketal and acetonide protected acetyl-carba(dethia)pantetheine. The acetyl-carba(dethia)pantetheine was liberated with the application of TFA at a pH of ~2 for a few hours in a water/methanol solvent then converted to the CoA with appropriate biosynthetic enzymes as previously reported. The acetyl-carba(dethia)-CoA product was purified by reverse-phase HPLC and confirmed with LCMS analysis against an authentic standard.

SaPC Mutagenesis, Expression and Purification.

All SaPC genes were expressed and purified in an identical manner. All reported sequence numbers are with respect to the S. aureus PC protein sequence and do not conform to the H. sapiens protein sequence numbering used in the protein databank for SaPC structures 3HO8 and 3BG5. Mutations in SaPC were generated using a whole-plasmid PCR technique according to the Agilent Quikchange II mutagenesis protocol, using the Bullseye Premium PR DNA Polymerase (Midwest Scientific; Fenton, MO). The K46A SaPC mutation was generated using forward primer 5'- GAT ATG CTG CAG ATG AAT CCT ATT TAG TTG -3' and reverse primer 5'- CAA CTA AAT AGG ATT CAT CTG CAG CAT ATC -3'. The E418A SaPC mutation was generated using forward primer 5'- CGA GCT ATG CGT ATT CGT GGT GTT AAA AC -3' and reverse primer 5'- GTT TTA ACA CCA CGA ATA CGC ATA GCT CG -3'. The R21A SaPC mutation was generated using forward primer 5'- CGT ATA TTC GCA GCG GCG GCA GAA TTA GAC -3' and reverse primer 5'- GTC TAA TTC TGC CGC CGC TGC GAA TAT ACG -3'. The K411A SaPC mutation was generated using forward primer 5'- GCG ATA TCA TTT GCG CAA GCA GAA GAA AAA -3' and reverse primer 5'- TTT TTC TTC TGC TTG CGC AAA TGA TAT CGC -3'. All constructs were confirmed by complete DNA sequencing of the entire gene.

All PC clones were co-transformed and co-expressed with the pCY216 vector encoding Escherichia coli biotin protein ligase A (BirA) to ensure complete biotinylation of the recombinantly expressed SaPC (19). Transformed E. coli BL21 (DE3) cells were cultured in 1× M9 minimal media, 100 μg/mL ampicillin, 30 μg/mL chloramphenicol, 0.5% w/v glycerol, 0.05% w/v glucose, 2 mM MgSO₄, and 0.25% w/v l-aspartate solution (20). Cultures contained 0.08 mg/mL l-leucine, 1× 18-amino acid solution and 1× trace metals as previously described (20). All cultures were grown at 37 °C to an optical density (OD₆₀₀) of 0.8 - 1.0, after which protein expression was induced by the addition of isopropyl 1-thio-β-D-galactopyranoside (IPTG) and l-arabinose to a final concentration of 0.5 mM and 25 mM, respectively. The culture was also supplemented with d-(+)-biotin at a final concentration of 3 mg/L. Induced cells were incubated at 16 °C for 16 - 24 hours before harvesting by centrifugation.

All SaPC enzymes were purified with sequential chitin affinity and anion exchange chromatography. Harvested cell paste (~20 – 30 g wet weight) was re-suspended in 200 mL binding buffer [20 mM HEPES, pH 7.8; 200 mM NaCl; 0.5 mM ethylenediaminetetraacetic acid (EDTA); 1 mM Tris (2-carboxyethyl) phosphine (TCEP); 1 mM phenylmethylsulfonyl flouride (PMSF); 1 μM pepstatin A; and 5 μM l-trans-3-carboxyoxiran-2-carbonyl- l-leucylagmatine (E-64)]. Cells were lysed by sonication for 8 minutes while maintaining the temperature in a range from 4 - 10 °C. The cellular debris was pelleted by centrifugation at 48000×g and the supernatant was loaded on a 10 mL chitin resin column (New England Biolabs, Ipswich, MA), with three re-applications of the column flow through. The column was washed with 100 mL of binding buffer and the resin was subsequently incubated overnight at 4 °C with 30 mL of cleavage buffer [20 mM HEPES, pH 7.8; 200 mM NaCl; 1 mM EDTA; and 50 mM dithiothreitol (DTT)] before collecting the liberated protein from the column. Two additional applications at 4 °C of 30 mL cleavage buffer (~4 hour incubations) were collected. The combined eluant from the three separate incubations were pooled and dialyzed overnight at 4 °C against a buffer compatible with anion-exchange chromatography (20 mM triethanolamine, pH 8.0; 50 mM NaCl; 1 mM EGTA; and 2 mM DTT). The dialyzed protein was loaded onto a 10 mL Q-Sepharose Fast Flow resin column (GE Healthcare), washed with 100 mL of dialysis buffer and eluted with a gradient from 50 mM to 1 M NaCl in dialysis buffer. SaPC typically eluted between 400 – 800 mM NaCl. The purified protein was pooled and dialyzed against storage buffer (20 mM HEPES, pH 7.2; 15 mM NaCl; 10 mM MgCl₂, and 1 mM TCEP) for two successive changes of 4 hours or more. The protein was concentrated to a final concentration of 10 – 25 mg/mL and was drop frozen in liquid nitrogen prior to storage at −80 °C. All protein concentrations were determined using the predicted extinction coefficient at 280 nm.

Pyruvate Carboxylation Assay.

Pyruvate carboxylation activity was measured by monitoring the change in absorbance at 340 nm resulting from the conversion of oxaloacetate to malate in the presence of the coupled enzyme, malate dehydrogenase (MDH). Reactions were performed in a buffer containing 0.1 M Tris-HCl (pH 7.8), 0.1 M KCl, and 7 mM MgCl₂. All substrates and coupling reagents were prepared as a 10× stock solution that resulted in final reaction concentrations of 25 mM NaHCO₃, 2.5 mM ATP, 0 – 4 mM acetyl-CoA, 12 mM sodium pyruvate, 0.12 mM NADH, and 10 U/mL MDH. The reaction was initiated by addition of the 10× stock substrate solution. The final SaPC concentration in the assay ranged from 2 – 30 μg/mL per reaction. All assays were performed in triplicate, and reported errors are the standard deviation resulting from three independent trials. The data were fit to Equation 1, which describes cooperative, non-essential allosteric activation at saturating substrate concentrations as a function of acetyl-CoA concentration ([A]), where $k_{\text{cat}}$ is the first-order rate constant in the presence of activator, $k_{\text{cat}}^{\circ }$ is the first-order rate constant in the absence of activator, $k_{\text{cat}}^{\text{app}}$ is the apparent first-order rate constant, $K_{\mathrm{a}}$ is the apparent activation constant, and $h$ is the Hill coefficient.

$$k_{cat}^{app}=\frac{k_{cat}^{°}+k_{cat}{[A]}^h}{K_a^h+{[A]}^h}$$ (Equation 1)

The titration data with 3’5’diphosphoadenosine were fit to equation 2, which describes allosteric activation and subsequent inhibition at saturating substrate concentrations, where $[A]$ is the concentration of 3’5’diphosphoadenosine, $k_{\text{cat}}$ is the first-order rate constant in the presence of activator, $k_{\text{cat}}^{\circ }$ is the first-order rate constant in the absence of activator, $k_{\text{cat}}^{\text{app}}$ is the apparent first-order rate constant, $K_{\mathrm{a}}$ is the apparent activation constant, $K_{\mathrm{i}}$ is the inhibition constant, and $h$ is the Hill coefficient. The data were best fit when the $K_{\mathrm{i}}$ was fixed at 5 mM and $k_{\text{cat}}^{\circ }$ was fixed at 1 s⁻¹ based on the experimentally determined value for $k_{\text{cat}}$ in the absence of activator.

$$k_{cat}^{app}=\frac{k_{cat}^{\circ }+{\mathrm{k}}_{\text{cat}}{\left(\frac{[\mathrm{A}]}{{\mathrm{K}}_{\mathrm{a}}}\right)}^{{}^{{}^{{}^{\mathrm{h}}}}}}{1+{\left(\frac{[A]}{K_a}\right)}^{{}^{{}^{{}^h}}}+{\left(\frac{[A]}{K_a}\right)}^{{}^{{}^{{}^h}}}\left(\frac{[A]}{K_i}\right)}$$ (Equation 2)

The titration data with l-aspartate were fit to equation 3, which describes allosteric inhibition at saturating substrate concentrations, where $k_{\text{cat}}$ is the first-order rate constant in the presence of inhibitor, $k_{\text{cat}}^{\circ }$ is the first-order rate constant in the absence of inhibitor, $k_{\text{cat}}^{\text{app}}$ is the apparent first-order rate constant, $K_{\mathrm{i}}$ is the inhibition constant, $h$ is the Hill coefficient, and $[\mathrm{I}]$ is the concentration of l-aspartate.

$$k_{cat}^{app}=\frac{k_{cat}^{°}+k_{cat}{[I]}^h}{K_i^h+{[I]}^h}$$ (Equation 3)

Pyruvate Carboxylation Inhibition Assays.

Pyruvate carboxylation activity was measured in the presence of CT domain inhibitors by monitoring the change in absorbance at 340 nm resulting from the conversion of oxaloacetate to malate in the presence of the coupled enzyme, MDH, as previously described (21). Reactions were performed in a buffer containing 0.1 M Tris-HCl (pH 7.8), 0.15 M KCl, 7 mM MgCl₂, 0.5% triton X-100, and 1% dimethyl sulfoxide (DMSO). All substrates and coupling reagents were prepared as a 10× stock solution, resulting in final reaction concentrations of 25 mM NaHCO₃, 2.5 mM ATP, 12 mM sodium pyruvate, 0.12 mM NADH, and 10 U/mL MDH. The enzyme was pre-incubated with the inhibitor and 250 μM acetyl-CoA in the presence of 5 μg/mL WT SaPC or 500 μM acetyl-CoA in the presence of 5 – 500 μg/mL for the SaPC mutants. The inhibitor concentrations ranged from 0 – 1 mM oxalate, 0 – 1 mM 2-hydroxy-3-(quinoline-2-yl)propenoic acid, or 0 – 50 mM phenylpyruvate. For all inhibitors, control reactions were run against MDH alone to confirm that the coupling enzyme was unaffected at the highest inhibitor concentration. The reaction was initiated by addition of the 10× stock substrate solution to the enzyme-acetyl-CoA-inhibitor incubation. All assays were performed in triplicate, and reported errors are the standard deviation resulting from three independent trials. The normalized velocity was calculated from equation 4, where $\mathrm{v}$ is the initial velocity, ${\mathrm{v}}_{\max }$ is the maximum velocity, and ${\mathrm{v}}_{\min }$ is the minimum velocity. The normalized velocities were then plotted against the concentration of inhibitor and the data was fit to equation 5 where ${\mathrm{v}}_{\max }$ represents the maximum normalized velocity, ${\mathrm{v}}_{\min }$ represents the minimum normalized velocity, $[I]$ represents the concentration of inhibitor and $h$ represents the Hill coefficient.

$$normalizedvelocity=\frac{(v^{\prime }-v_{min})}{(v_{max}-v_{min})}$$ (Equation 4)

$$normalizedvelocity=v_{max}+\left(\frac{v_{min}-v_{max}}{1+{\left(\frac{{IC}_{50}}{[I]}\right)}^{{}^{{}^{{}^h}}}}\right)$$ (Equation 5)

ADP Phosphorylation Assay.

ADP phosphorylation activity of PC was measured spectrophotometrically at 340 nm by following the conversion of glucose to glucose-6-phosphate and, subsequently, to 6-phosphogluconate using the coupling enzymes hexokinase and glucose-6-phosphate dehydrogenase, as previously described (22). Reactions were performed in a buffer containing 100 mM Tris-HCl (pH 7.8), 0.1 M KCl, and 7.5 mM MgCl₂. WT SaPC was used to initiate the reaction, at a final concentration of 30 μg/mL. The reactions included 20 mM carbamoyl phosphate, 0.24 mM NADP⁺, 3 mM ADP, 200 μM glucose, 10 U/mL of glucose-6-phosphate dehydrogenase, 2 U/mL hexokinase, and 0 – 8 mM acetyl-CoA. All assays were performed in triplicate, and reported errors are the standard deviation resulting from three independent trials.

Size Exclusion Chromatography.

Size exclusion chromatography was performed on a BioRad NGC Quest 10 Plus Chromatography system using a Superose 6 10/300 GL size exclusion column from Amersham Pharmacia Biotech (Piscataway, NJ). WT and mutant SaPC samples, both at a concentration of 1 mg/mL, were loaded onto the chromatography column with a volume of 1 mL. The protein was eluted using a buffer containing 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 2 mM TCEP at a flow rate of 0.3 mL/min. The eluted samples were monitored at 280 nm. The molecular weights and multimerization states of tetrameric WT and dimeric ΔBCΔBCCP PC were previously determined (23).

Protein Crystallization.

Crystals of R21A SaPC were obtained at room temperature by the batch crystallization method. The drop consisted of 13.3% MEPEG 2K, 133 mM KNO₃⁻, 50 mM 4-morpholinepropanesulfonic acid (MOPS; pH 7.0), 1.5% v/v 2-propanol, 0.625 mM acetyl-CoA, 1.25 mM adenosine triphosphate (ATP), and 9.65 mg/mL R21A SaPC. The drop volume was 10 μL. Under these conditions, crystals emerged spontaneously and grew to full size (~0.25 mm³) after 2 – 4 days, with a cube-like morphology. Crystals were transferred, via a quartz capillary, to a synthetic stabilizing solution that consisted of 14.6% MEPEG 2K, 147 mM KNO₃⁻, 50 mM MOPS (pH 7.0), 1.5% v/v 2-propanol, 1.4 mM ATP, and 0.69 mM acetyl-CoA. Crystals were incubated in the stabilizing solution for 5 minutes prior to being incrementally transferred into a cryoprotectant solution that consisted of 14.6% MEPEG 2K, 147 mM KNO₃⁻, 50 mM MOPS pH 7.0, 1.5% v/v 2-propanol, 1.4 mM ATP, 0.69 mM acetyl-CoA, and 20% v/v ethylene glycol. The crystals were successively transferred 5 times into solutions with increasing amounts of cryoprotectant.

Data Collection, Structure Determination, and Refinement.

X-ray diffraction data were collected at the Life Sciences Collaborative Access Team beamline 21-ID-F at the Advanced Photon Source, Argonne National Laboratory, on a Rayonix MX-300 s/n 023 detector with an X-ray wavelength of 0.97872 Å. A total of 360 diffraction images were collected at an oscillation angle of 1° (360° total) for an exposure time of 2 s, with a detector distance of 300 mm. The images were processed using autoPROC (24), which merged and scaled the anisotropic diffraction data in AIMLESS (25), using the programs TRUNCATE and UNIQUE (26,27) to determine the best estimate of the resolution limit. The phases were determined by molecular replacement using the program Phaser (28) with the search ensembles consisting of four copies of the AD+CT domain (3HO8; residues 493-1092), followed by four copies of the BC domain (3HO8; residues 1 – 492). The resulting model was put through multiple rounds of manual model building using COOT (29), and refinement using Phenix.Refine (30). Refinements were performed by maintaining non-crystallographic symmetry restraints in the early rounds of refinement and including TLS parameters in all refinements. Water molecules were added using COOT and Phenix.Refine, with manual verification. The refined ligand coordinates for acetyl-CoA and CoA were optimized using eLBOW (electronic ligand building and optimization workbench) (31). Data collection and processing statistics are listed in Table S1.

RESULTS

Probing the acetyl moiety binding site with non-hydrolysable acetyl-CoA analogs.

In pyruvate carboxylase, the acetyl moiety is known to contribute several orders of magnitude to the activation constant for acetyl-CoA, but the molecular basis for this phenomenon has not been fully described. Efforts to define the role of the acetyl moiety have been impeded by the lack of electron density corresponding to the acetyl moiety in co-crystal structures of PC with acetyl-CoA. Presumably, this is a consequence of either enzyme-mediated or spontaneous hydrolysis of acetyl-CoA (32). To better define the contributions of the acetyl moiety to the allosteric activation of PC by acetyl-CoA, non-hydrolysable/slowly hydrolysable dethia analogs of acetyl-CoA [acetyl-carba(dethia)-CoA, acetyl-oxa(dethia)-CoA, and acetyl-aza(dethia)-CoA (Figure S1)] were synthesized and assessed for their ability to activate SaPC. The activation of SaPC by each of these analogs was evaluated for both the overall degree of activation $(k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ })$ and the apparent activation constant $(K_{\mathrm{a}})$. All three analogs had the same degree of activation as acetyl-CoA but displayed elevated apparent $K_{\mathrm{a}}$ values relative to acetyl-CoA (Figure 2A). Both acetyl-carba(dethia)-CoA and acetyl-oxa(dethia)-CoA had apparent $K_{\mathrm{a}}$ values comparable to CoA, while acetyl-aza(dethia)-CoA had an apparent $K_{\mathrm{a}}$ value that was two orders of magnitude higher than CoA (12,33). This indicates that the thioester linkage plays a key role in positioning the acetyl moiety of acetyl-CoA in the allosteric site. In addition, 3’5’-diphosphoadenosine, which lacks the S-acetyl phosphopantetheine arm of acetyl-CoA, was assessed for activation of SaPC. At concentrations up to 7 mM, 3’5’-diphosphoadenosine activated SaPC, but it was inhibitory at concentrations greater than 7 mM (Figure S2). The inhibitory effect at high concentrations of 3’5’-diphosphoadenosine was attributed to competitive inhibition with ATP in the BC domain active site. A similar phenomenon has been observed with other nucleotide analogues (34,35). A fit of the 3’5’-diphosphoadenosine titration data to Equation 2, which is derived from a kinetic scheme for non-essential activator inhibition, yielded an estimated apparent $K_{\mathrm{a}}$ value of ~5.9 mM, which is three orders of magnitude greater than the apparent $K_{\mathrm{a}}$ values of the acetyl-CoA analogs (Table 1). These data confirm that the acetyl thioester makes a critical contribution to the apparent binding affinity of acetyl-CoA, but that this linkage is not required to achieve full activation of the enzyme.

Figure 2. Acetyl-CoA analogs activate SaPC.

A. The degree of activation $(k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ })$ for the carboxylation of pyruvate by WT SaPC, titrated with acetyl-CoA (black), CoA (blue), acetyl-oxa(dethia)-CoA (green), acetyl-carba(dethia)-CoA (purple), and acetyl-aza(dethia)-CoA (red). The $k_{\text{cat}}^{\circ }$ value for WT SaPC was determined to be 0.94 s⁻¹ in the absence of acetyl-CoA. B. The degree of activation $(k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ })$ for the carboxylation of pyruvate by WT SaPC in the presence of 20 mM l-aspartate titrated with acetyl-CoA (black), CoA (blue) and acetyl-carba(dethia)-CoA (purple), acetyl-oxa(dethia)-CoA (green), and acetyl-aza(dethia)-CoA (red). The black lines represent the fit of the data to Equation 1. Error bars represent standard deviations from three independent measurements.

Table 1.

Activation of the WT SaPC pyruvate carboxylation reaction by acyl CoAs.

		Acyl-CoA			Acyl-CoA + 20 mM l-aspartate
Compound	$k_{\text{cat}}^{\circ }({\mathrm{s}}^{-1})$	$K_{\mathrm{a}}(\mu \mathrm{M})$	$h$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$	$K_{\mathrm{a}}(\mu \mathrm{M})$	$h$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$
	0.94 ± 0.09
Acetyl-CoA		1.5 ± 0.1	1.5 ± 0.1	13 ± 1	4.1 ± 0.3	3.3 ± 0.6	14 ± 1
CoA		40 ± 3	1.4 ± 0.2	14 ± 1	130 ± 10	1.8 ± 0.2	16 ± 1
Acetyl-carba(dethia)-CoA		53 ± 4	1.5 ± 0.2	13 ± 1	190 ± 10	3.1 ± 0.6	16 ± 1
Acetyl-oxa(dethia)-CoA		73 ± 11	1.5 ± 0.3	13 ± 1	270 ± 20	3.3 ± 0.7	16 ± 1
Acetyl-aza(dethia)-CoA		2500 ± 1000	1.3 ± 0.4	15 ± 2	ND*	ND	ND
3′5′ diphosphoadenosine		5900 ± 3100	1.6 ± 0.7	5.8 ± 0.8	ND	ND	ND

^*ND = could not be determined

The dethia acetyl-CoA analogs were also assessed for their ability to overcome allosteric inhibition by l-aspartate in the pyruvate carboxylation assay. l-Aspartate, an allosteric inhibitor of PC, is competitive with respect to acetyl-CoA in various bacterial and fungal PC enzymes (16,17). To assess whether the dethia analogs could displace l-aspartate, they were titrated in the presence of a fixed concentration of 20 mM l-aspartate. In the presence of l-aspartate, the acetyl-carba(dethia)-CoA and acetyl-oxa(dethia)-CoA analogs behaved much like acetyl-CoA and CoA: all of these activators maintained the same degree of activation and the apparent $K_{\mathrm{a}}$ values increased ~4-fold in the presence of l-aspartate. Notably, the Hill coefficients $(h)$ for acetyl-CoA, acetyl-carba(dethia)-CoA and acetyl-oxa(dethia)-CoA all increased in the presence of l-aspartate (Table 1) consistent with displacement of l-aspartate (17). Interestingly, unlike the other activators, CoA did not exhibit a substantially increased Hill coefficient in the presence of l-aspartate (Table 1). This suggests that CoA is unable to fully displace l-aspartate, but that allosteric activation by CoA fully overcomes the allosteric inhibition by l-aspartate. Finally, acetyl-aza(dethia) CoA, which exhibits a much higher apparent $K_{\mathrm{a}}$ value compared to the other analogs, was unable to overcome l-aspartate inhibition, indicating that it was unable to displace l-aspartate (Figure 2B).

Characterization of BC dimer interface residues at the acetyl moiety binding site.

To identify and characterize residues that interact with the acetyl moiety of acetyl-CoA, a series of residues at the BC dimer interface were mutated to alanine. In X-ray crystal structures of SaPC, residues Lys₄₆, Glu₄₁₈, and Arg₂₁ are located at the BC dimer interface, and in close proximity to the thiol of CoA (Figure 3A). These residues are conserved among PC enzymes but are not conserved in the larger family of biotin carboxylases (Figure 3B). Mutation of Lys₄₆, Glu₄₁₈, and Arg₂₁ to alanine was designed to specifically probe the potential for the BC dimer interface to mediate the allosteric effect of acetyl-CoA (Figure 3C). Mutation of Lys₄₁₁ to alanine was designed as a control, to perturb the BC dimer interface at a site distant (~12 Å) from the putative acetyl moiety binding site.

Figure 3. The acetyl moiety binding site at the BC dimer interface.

A. Residues within a 10 Å radius of the thiol of CoA in SaPC (PDB ID 3HO8) were candidates for mutagenesis. The distance from the thiol of CoA in chain A is denoted for each residue. The light blue shading denotes residues originating from one subunit (chain A) while the darker blue shading denotes residues originating from the second subunit (chain C). B. Multiple sequence alignment for PC from various organisms and other biotin dependent carboxylases: biotin carboxylase (BC), acetyl coenzyme A carboxylase (ACC), urea carboxylase (UC), 3-methylcrotonyl coenzyme A carboxylase (MCC), and propionyl coenzyme A carboxylase (PCC). The asterisk represents conserved residues in PC that are within 10 Å of the thiol of CoA, highlighted in panel A. C. The BC dimer interface highlighting the residues targeted for mutagenesis. Inter-subunit hydrogen bonding and ionic interactions are represented by a dotted line and the bond distances are denoted next to each bond. The dimer interface is represented by a thick blue line. D. Elution profiles for K46A SaPC (blue), E418A SaPC (red), R21A SaPC (green), and K411A SaPC (purple). The elution profile for each mutant is compared to the elution profile for tetrameric WT SaPC (black), and dimeric ΔBCΔBCCP SaPC (brown).

Since the mutated residues were located at the BC dimer interface, the impact of the mutations on the oligomerization state of the mutant enzymes was first assessed. Analytical size exclusion chromatography was performed for WT SaPC, ΔBCΔBCCP SaPC and each of the BC interface mutants (Figure 3D). WT SaPC ran as a tetramer while the ΔBCΔBCCP SaPC construct, which lacks both the BC and BCCP domains, ran as a dimer as previously demonstrated for the identical construct in Rhizobium etli PC (23). The R21A SaPC mutant had the same elution profile as WT SaPC, while the other mutations demonstrated a small increase in the percentage of dimer. Nevertheless, in each case, the mutated construct populations favored the tetrameric species and did not differ significantly from the profile of the WT enzyme (Figure 3D).

The mutations K46A, E418A, R21A and K411A were kinetically characterized in the pyruvate carboxylation reaction (Table 2, Figure 4). The K411A mutation was employed as a control to account for perturbations at the BC dimer interface that are independent of direct interactions with acetyl-CoA. All the mutants retained pyruvate carboxylation activity in the presence of both acetyl-CoA and CoA but the response of each mutant to activation by acetyl-CoA and CoA differed compared to WT SaPC. Relative to WT SaPC, the K46A SaPC mutant exhibited a reduced degree of activation and a substantially elevated apparent $K_{\mathrm{a}}$ value for both acetyl-CoA and CoA. The R21A SaPC mutant also exhibited a substantially elevated apparent $K_{\mathrm{a}}$ value for acetyl-CoA and CoA, but both allosteric effectors were able to activate the R21A mutant to the same degree as WT SaPC (~14-fold). Of all the mutants surveyed, the K411A SaPC mutation, located distant from the acetyl-CoA binding site, displayed kinetic constants most like WT. This mutant exhibited only a slightly elevated apparent $K_{\mathrm{a}}$ value relative to WT SaPC and exhibited no change in the degree of activation relative to WT SaPC. The most dramatic effect was observed in the E418A SaPC mutant. This mutant was unresponsive to activation by both acetyl-CoA and CoA, even at extremely high concentrations of the activators (up to 4 mM acetyl-CoA and 30 mM CoA) (Figure 4A).

Table 2.

Activation of the pyruvate carboxylation reaction by acetyl-CoA and CoA for SaPC mutations in the BC dimer interface.

			Acetyl-CoA		CoA
Enzyme	$k_{\text{cat}}^{\circ }({\mathrm{s}}^{-1})$	% WT Activity	$K_{\mathrm{a}}(\mu \mathrm{M})$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$	$K_{\mathrm{a}}(\mu \mathrm{M})$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$	ratio
WT SaPC	1.5 ± 0.1	N/A	1.5 ± 0.1	12 ± 1	49 ± 7	13 ± 1	33
K46A SaPC	0.45 ± 0.02	31%	340 ± 30	7.9 ± 0.3	1500 ± 100	5.5 ± 0.3	4
E418A SaPC	0.039 ± 0.001	3%	ND	ND	ND	ND	ND
R21A SaPC	0.089 ± 0.004	6%	340 ± 50	14 ± 1	1800 ± 53	14 ± 0.27	5
K411A SaPC	0.23 ± 0.01	16%	5.8 ± 0.2	14 ± 1	100 ± 4.8	14 ± 0.44	17

^*ND = could not be determined

Figure 4. Kinetic characterization of BC dimer interface mutations.

A. The degree of activation $(k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ })$ for the carboxylation of pyruvate as a function of acetyl-CoA (solid line, filled circles) and CoA (dashed line, open circles) concentrations for WT SaPC (black), K411A SaPC (purple), R21A SaPC (green), K46A SaPC (blue), and E418A SaPC (red). Data were fit to Equation 1. B. The degree of activation $(k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ })$ for the phosphorylation of ADP as a function of acetyl-CoA concentration for WT SaPC (black), K411A SaPC (purple), R21A SaPC (green), K46A SaPC (blue), and E418A SaPC (red). K411A SaPC and WT SaPC data were fit to Equation 1. R21A SaPC, K46A SaPC and E418A SaPC were fit to a linear regression. C. Pyruvate carboxylation activity, plotted as $k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$, as a function of l-aspartate for WT SaPC (black), K411A SaPC (purple), R21A SaPC (green), K46A SaPC (blue), and E418A SaPC (red). K411A SaPC and WT SaPC data were fit to Equation 3. R21A SaPC, K46A SaPC and E418A SaPC were fit to a linear regression. Error bars are representative of the standard deviation from 3 independent measurements.

To determine the impact of these mutations on the catalytic activity of the BC domain, the mutants were assessed for the ability of acetyl-CoA to activate the phosphorylation of ADP by carbamoyl phosphate, a well-established assay of the isolated BC domain half reaction (Figure 4B, Table 3). It was determined that, for each of the mutated SaPC enzymes, 50 mM carbamoyl phosphate was saturating (Figure S3), and this concentration of carbamoyl phosphate was maintained for all titrations with acetyl-CoA. The rate of ADP phosphorylation was reduced for all the mutants, with E418A SaPC having the lowest relative rate. Surprisingly, K46A SaPC, E418A SaPC and R21A SaPC were all unable to be activated by acetyl-CoA in the BC domain half reaction, up to 1 mM acetyl-CoA. K411A SaPC responded to acetyl-CoA but was not activated to the same degree as WT SaPC and had an elevated apparent $K_{\mathrm{a}}$ value. In all cases, including WT, there was slight inhibition at concentrations of acetyl-CoA greater than 4 mM (Figure S4).

Table 3.

Activation of the ADP phosphorylation reaction by acetyl-CoA and CoA for SaPC mutations in the BC dimer interface.

			Acetyl-CoA		CoA
Enzyme	$k_{\text{cat}}^{\circ }({\mathrm{s}}^{-1})\times {10}^{-2}$	% WT Activity	$K_{\mathrm{a}}(\mu \mathrm{M})$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$	$K_{\mathrm{a}}(\mu \mathrm{M})$	$k_{\text{cat}}^{\text{app}}∕k_{\text{cat}}^{\circ }$
WT SaPC	1.9 ± 0.1	N/A	15 ± 6	4.8 ± 0.7	320 ± 30	2.2 ± 0.1
K46A SaPC	0.35 ± 0.01	18%	ND*	ND	ND	ND
E418A SaPC	0.18 ± 0.21	9%	ND	ND	ND	ND
R21A SaPC	0.43 ± 0.01	22%	ND	ND	ND	ND
K411A SaPC	1.4 ± 0.1	74%	14 ± 1	2.6 ± 0.1	420 ± 30	1.9 ± 0.1

^*ND = could not be determined

The precise binding site for the allosteric inhibitor, l-aspartate, is unknown, but it has been shown to be mutually exclusive with acetyl-CoA, strongly suggesting that the allosteric sites may overlap (36). To assess whether BC interface mutations near the acetyl moiety binding site could also impact l-aspartate inhibition, we titrated l-aspartate in the pyruvate carboxylation reaction. For WT SaPC, we observed allosteric inhibition as previously described (17). WT SaPC and K411A SaPC were fully inhibited with $K_{\mathrm{i}}$ values of 18 ± 1 and 24 ± 3 mM, respectively. However, K46A SaPC, E418A SaPC and R21A SaPC were not inhibited by l-aspartate up to 100 mM (Figure 4C). Inhibition was observed in all mutants at concentrations exceeding 100 mM l-aspartate (Figure S5). However, l-glutamate, which is not an inhibitor of bacterial PC, also inhibited WT SaPC at concentrations greater than 100 mM, indicating that inhibition at these highly elevated concentrations is a nonspecific effect (Figure S5).

Co-crystal structure of R21A SaPC with acetyl-CoA.

To further characterize the role of the BC dimer interface on allostery in PC, we sought to structurally characterize the BC dimer interface mutants in the presence of acetyl-CoA. Only R21A SaPC, grown in the presence of 0.69 mM acetyl-CoA and 1.4 mM ATP, yielded high quality crystals that diffracted to 2.7 Å resolution. The crystals belonged to space group P21 (a = 94 Å, b = 254 Å, c = 126 Å, α = 90°, β = 111°, γ = 90°), with 4 subunits in the asymmetric unit. The resulting omit maps revealed strong difference density for acetyl-CoA in chain A (Figure 5). Chain C also had strong difference density for coenzyme A, but this was best modeled by the hydrolyzed CoA thiol rather than the intact acetyl-CoA thioester. The omit maps for acetyl-CoA and CoA are best contoured at 2 σ, but polder maps generated for both acetyl-CoA in chain A, and CoA in chain C, strongly support the modeled position and orientation of these ligands (Figure S6). All four subunits revealed BCCP-biotin located in the so-called “exo binding site” at the C-terminal end of the AD (33,37). Much of the BCCP domain was disordered and could not be modeled. Crystal contacts stabilized the B-subdomain lid in an open conformation in the BC domains of chains A and B. In chains C and D, the B subdomain lid was fully disordered and could not be modeled. The overall structure of R21A SaPC is similar to the structure of WT SaPC in complex with CoA (PDB ID 3HO8, Figure S7). The BC domains of chain A for 3HO8 and the R21A SaPC structure aligned very well, with a rmsd of 0.6 Å for all mainchain atoms. When these subunits were aligned, the other subunit of the dimer also remained well aligned, with a rmsd of 1.4 Å for all mainchain atoms (Figure S7). This demonstrates a very similar BC dimer interface in both structures, reinforcing that the BC dimer interface in R21A SaPC is intact and that neither the mutation nor the binding of acetyl-CoA significantly alters the dimer interface in SaPC. Within the larger tetramer context, there are variations in the precise arrangement of individual subunits compared to WT SaPC. However, it is worth noting that structural variances in the tetramer are frequently observed in PC structures and have been described in detail elsewhere (12,33,37-40).

Figure 5. R21A SaPC in complex with acetyl-CoA.

A. Omit mF_O-DF_C electron density maps contoured at 2 σ (green mesh) corresponding to acetyl-CoA in chain A. CoA from chain C (white) is overlaid with acetyl-CoA (grey) from chain A. B. Omit mF_O-DF_C electron density maps contoured at 2 σ (green mesh) corresponding to CoA in chain C. Acetyl-CoA from chain A (grey) is overlaid with CoA (white) from chain C. C. Acetyl-CoA from chain A with mF_O-DF_C electron density maps contoured at 3 σ (green mesh) and 2mF_O-DF_C electron density contoured at 1 σ (blue mesh) surrounding residues Lys₄₆, from chain C, and Glu₄₁₈, from chain A. The distances from the acetyl moiety to Lys₄₆ and Glu₄₁₈ are displayed. The water that occupies the space typically occupied by Arg₂₁ in WT SaPC is shown as a red sphere.

Unlike many other crystal structures that have weak electron density corresponding to the acetyl phosphopantetheine arm, the structure of R21A SaPC revealed relatively strong difference density in chain A for the entire acetyl-CoA molecule, including the acetyl moiety (12,41). Acetyl-CoA in chain A, and CoA in chain B, have their nucleotidyl portions in the same orientation, but the density in chain A extends further to encompass the acetyl moiety (Figure 5A and B). The position and orientation of acetyl-CoA in chain A is very similar to the acetyl-CoA molecules modeled in recent cryo EM structures of Lactococcus lactis PC and Homo sapiens PC (37,42) (Figure S8) as well as to previously published structures of SaPC in complex with CoA. Acetyl-CoA is stabilized by a well-characterized set of interactions between Arg₄₂₀, Arg₄₂₂ and Arg₄₆₅ and the phosphates extending from the nucleotidyl portion of acetyl-CoA (13). In addition to their stabilizing interactions at the dimer interface, Lys₄₆, Glu₄₁₈ and, likely, Arg₂₁, also interact with the acetyl moiety of acetyl-CoA (Figure 5C). The density for the carbonyl oxygen of the acetyl moiety was modeled to position the oxygen just 3.4 Å from Glu₄₁₈, the residue at the BC dimer interface that is most critical to mediating the allosteric effect of acetyl-CoA (Figure 5C and Figure 4A). Lys₄₆, which also contributes significantly to mediating the allosteric effect, is located ~4.5 Å from the modeled position of the carbonyl oxygen. It must be noted that both the K46A and R21A SaPC mutants were much less discriminating in their relative activation by acetyl-CoA compared to CoA, with only a ~4 – 5-fold reduction in apparent $K_{\mathrm{a}}$ on going from acetyl-CoA to CoA, compared to ~30-fold for WT SaPC (Table 2). This reduced discrimination for acetyl-CoA may have contributed to the success in capturing acetyl-CoA in the R21A structure, but it also introduces a caveat: the lack of discrimination for acetyl-CoA in R21A SaPC suggests that this structure may not represent the true orientation of acetyl-CoA in the WT enzyme.

Acetyl-CoA alters CT domain inhibition.

To investigate the impact of acetyl-CoA and the BC dimer interface on long range communication between the BC and CT domains, we assessed the influence of acetyl-CoA on the inhibition of PC by a well-characterized set of CT domain inhibitors. Oxalate is known to bind in the CT domain with a similar binding orientation to pyruvate and is a potent inhibitor of PC (22,43,44). Interestingly, the IC₅₀ value for oxalate inhibition of WT SaPC increased as a function of acetyl-CoA concentration, resulting in more than a 20-fold increase in the IC₅₀ at saturating concentrations of acetyl-CoA (Figure 6A, Table S2). A similar phenomenon was observed for two other competitive inhibitors with respect to pyruvate, 2-hydroxy-3-(quinoline-2-yl)propenoic acid and phenylpyruvate (21) (Figure S9, Table S3). CoA also increased the IC₅₀ value of oxalate ~13-fold, indicating that this phenomenon is a general feature of allosteric activation and is not specific to the acetyl moiety (Figure 6B, Table S2). Moreover, the allosteric inhibitor, l-aspartate, did not alter the IC₅₀ value of oxalate (Figure 6B, Table S2). The ability of CoA, and the inability of l-aspartate, to alter the IC₅₀ value of oxalate suggests that the BC dimer interface does not contribute to reduced CT domain inhibition in the presence of acetyl-CoA. To test this, we examined the oxalate inhibition of both R21A and K46A in the presence and absence of 2.5 mM acetyl-CoA, which is equivalent to a practical upper limit of ~7 times the $K_{\mathrm{a}}$ value for both mutants. At 2.5 mM acetyl-CoA, both the R21A and K46A mutations displayed an 8-fold increase in the IC₅₀ value for oxalate, equivalent to the ~6-fold increase in the IC₅₀ value for oxalate inhibition of WT SaPC at ~7 times the $K_{\mathrm{a}}$ value (Figure 6C, Table S4). Although E418A SaPC retained only 3% of WT SaPC activity, it was inhibited by all three inhibitors (oxalate, 2-hydroxy-3-(quinoline-2-yl)propenoic acid, and phenylpyruvate). This inhibition was unchanged in the presence of 500 μM acetyl-CoA, but this is consistent with the inability of E418A to respond to activation by acetyl-CoA. K411A SaPC also exhibited a substantially elevated IC₅₀ value for CT domain inhibitors in the presence of acetyl-CoA (Figure S10, Table S3). Taken together, these results indicate that, while acetyl-CoA alters inhibition in the CT domain, the acetyl moiety binding site does not contribute to this phenomenon.

Figure 6. The effect of effector molecules and mutations on oxalate inhibition.

A. Normalized velocities for the pyruvate carboxylation reaction catalyzed by WT SaPC at increasing concentrations of oxalate in the absence of acetyl-CoA (open circles, dashed line), in the presence of 2.5 μM acetyl-CoA (squares), 5 μM acetyl-CoA (triangles), 10 μM acetyl-CoA (inverted triangles), and 250 μM acetyl-CoA (diamonds). B. Normalized velocities for the pyruvate carboxylation reaction catalyzed by WT SaPC at increasing concentrations of oxalate in the absence of acetyl-CoA (open circles, dashed line), in the presence of 5 mM l-aspartate (triangles), and in the presence of 500 μM CoA (squares). C. Normalized velocities for the pyruvate carboxylation reaction at increasing concentrations of oxalate. WT SaPC is shown in black in the absence of acetyl-CoA (open circles, dashed line) and in the presence of 10 μM acetyl-CoA (closed circles, solid line). R21A SaPC (green) in the absence of acetyl-CoA (open circles, dashed line) and in the presence of 2.5 mM acetyl-CoA (closed circles). K46A SaPC (blue) in the absence of acetyl-CoA (open circles, dashed line) and in the presence of 2.5 mM acetyl-CoA (closed circles). Normalized velocities were calculated using Equation 4 and the data were fit to Equation 5, where ${\mathrm{v}}_{\max }$ was constrained to 1 and ${\mathrm{v}}_{\min }$ was constrained to 0. Error bars, shown in red, represent the propagation of error from three separate measurements.

DISCUSSION

The allosteric regulation of PC by acetyl-CoA and l-aspartate has been known for over 60 years (36,45), but a complete description of the binding and mechanism of acetyl-CoA activation and l-aspartate inhibition remains an active area of investigation. Acetyl-CoA binds at the allosteric domain and extends into the BC dimer interface, but the binding site for the acetyl moiety has not been defined (13). The precise binding site for the allosteric inhibitor, l-aspartate, is also unknown. Acetyl-CoA and l-aspartate binding is mutually exclusive and, therefore, l-aspartate has been postulated to bind at, or near, the acetyl-CoA binding site (17). Our structural and kinetic data more precisely define the binding site for the acetyl moiety at the BC dimer interface and indicate that this also represents the overlapping binding site for the allosteric inhibitor, l-aspartate. We conclude that the acetyl moiety binding site at the BC dimer interface of SaPC represents an allosteric “hot spot” on the enzyme.

Prior studies probing the acetyl-CoA binding site have primarily focused on residues positioned near the ribonucleotide of acetyl-CoA (13,46,47). The 3’5’-ADP portion of acetyl-CoA binds at the allosteric domain and is both necessary and sufficient for activation since 3’5’-diphosphoadenosine activates SaPC (Figure S2), while acetylpantetheine, which lacks the nucleotidyl portion, binds to but does not activate PC (48). In addition to the essential role of the nucleotidyl moiety in acetyl-CoA activation, the acetyl moiety contributes significantly to the binding affinity, increasing it ~80 fold in SaPC (33). In this study, dethia acetyl-CoA analogs [acetyl-carba(dethia)-CoA, acetyl-oxa(dethia)-CoA, and acetyl-aza(dethia)-CoA] fully activated WT SaPC, but with 100 – 1000-fold lower binding affinities than acetyl-CoA (Figure 2A). Since CoA also has a ~100-fold lower affinity than acetyl-CoA, we conclude that the reduced binding affinities of the dethia analogs reflect stringent requirements on the intermediate polarity/hydrophobicity of the thioester linkage rather than a requirement for the sulfur atom. Acetyl-aza(dethia)-CoA had low binding affinity, with an apparent $K_{\mathrm{a}}$ value equivalent to that measured for 3’,5’-diphosphoadenosine. Notably, acetyl-aza(dethia)-CoA was also unable to activate WT SaPC in the presence of l-aspartate (Figure 2B and Table 1). Acetyl-aza(dethia)-CoA, therefore, likely does not extend its N-acetyl-ethylenediamine group into the BC dimer interface due to the potential of the amide to serve as a strong hydrogen bond donor. These results are consistent with an overlapping binding site for acetyl-CoA and l-aspartate and suggest that the acetyl moiety binding site is the most likely site of overlap with l-aspartate.

The protein-ligand interactions at the BC dimer interface and near to the acetyl moiety binding site for acetyl-CoA have not previously been assessed for their role in communicating the allosteric effect. Residues Arg₂₁, Lys₄₆, and Glu₄₁₈ were mutated to alanine due to their location in the BC dimer interface, their proximity to the acetyl moiety binding site, and their sequence conservation. Lys₄₆ of SaPC is of particular interest because it is conserved in nearly all PC enzymes with the notable exception of PC from Corynebacterium glutamicum, which is not subject to allosteric activation by acetyl-CoA (49,50). In this study, we also employed K411A as a control to ensure that observed changes in allosteric activation are not merely due to disruptions of interactions at the BC dimer interface. While Lys₄₁₁ is located at the BC dimer interface, it is distant from the acetyl moiety binding site. The K411A mutation did not substantially perturb the kinetic behavior of SaPC, in stark contrast to mutations of BC dimer interface residues located near the acetyl moiety.

The goal of our mutagenic study was to disrupt the protein-ligand interactions while keeping the BC dimer interface intact. All the mutants retained their ability to form tetramers and maintained catalytic activity (Figure 3D). The observed kinetic differences in the mutant constructs, therefore, are not simply attributed to structural perturbations at the dimer interface. Rather, the altered kinetics arise from the mutation of residues that are directly responsible for conveying the allosteric effect. Moreover, the oligomerization state of the R21A SaPC mutation was unaltered relative to the WT enzyme (Figure 3D) and the R21A SaPC crystal structure showed no significant structural deviations when compared to the structure of WT SaPC (Figure 5 and Figure S7). Residues at the BC dimer interface have previously been subjected to mutagenesis in SaPC, but these mutations were designed to fully disrupt the BC dimer interface through charge non-complementarity and, in all cases, resulted in inactive protein (51). The retention of activity in the current series of mutants, therefore, is further evidence of the structural integrity of the mutated tetramers and allows for a more careful study of their role in mediating allosteric interactions.

Both the R21A SaPC crystal structure and the kinetics of the mutated constructs implicate three residues at the acetyl moiety binding site (Arg₂₁, Lys₄₆, and Glu₄₁₈) in the allosteric mechanism. The crystal structure of R21A SaPC reveals the interactions between these residues and acetyl-CoA (Figure 5). This is the first time that an X-ray crystal structure of PC has revealed an unhydrolyzed molecule of acetyl-CoA at the allosteric site. Glu₄₁₈ is particularly noteworthy: while the E418A mutant retained a low level of catalytic activity, it was not allosterically activated by either acetyl-CoA or CoA, highlighting its critical role in mediating the allosteric mechanism (Figure 4A and B). In comparison, K46A displayed a reduced degree of activation, while R21A could be fully activated by both acetyl-CoA and CoA in the pyruvate carboxylation reaction. Both R21A and K46A had greatly reduced apparent binding affinities for acetyl-CoA, as indicated by their elevated apparent $K_{\mathrm{a}}$ values. Moreover, none of the mutants near the acetyl moiety binding site could be activated by acetyl-CoA in the assay of the BC domain-catalyzed ADP phosphorylation reaction (Figure 4B). The carboxyl moiety of Glu₄₁₈ is <3.5 Å from the carbonyl of the acetyl moiety of acetyl-CoA in chain A of the R21A SaPC structure and forms ionic interactions with Arg₂₁ and Lys₄₆ across the BC dimer interface (Figure 3C and D). Consequently, Glu₄₁₈ is ideally positioned to communicate the effect of acetyl-CoA binding across the BC dimer interface. The complete loss of allosteric regulation by both acetyl-CoA and l-aspartate for the E418A mutant suggests a direct and essential role for this residue in communicating the allosteric effect across the dimer interface. Lys₄₆ likely serves a similar purpose but acts as a more general transducer to facilitate allosteric activation from the 3’5’-ADP portion of acetyl-CoA. The N_ε of Lys₄₆ projects deep into the BC dimer interface near the acetyl moiety of acetyl-CoA, while residing at the furthest extension of loop 44 – 50 that wraps around the adenosyl moiety of acetyl-CoA. Consequently, Lys₄₆ is positioned to contribute multiple molecular interactions with the nucleotide base and to translate this binding interaction directly into the BC dimer interface.

The allosteric inhibitor, l-aspartate, binds cooperatively to PC and is competitive with respect to acetyl-CoA (16,17), but the binding site for l-aspartate is unknown. Previous mutagenic studies conducted on residues surrounding the nucleotidyl portion of acetyl-CoA had no effect on l-aspartate inhibition (13,17). In this study, R21A, K46A and E418A SaPC all lost the ability to be inhibited by l-aspartate (Figure 4C and Figure S5) while the control BC dimer mutation, K411A SaPC, was inhibited to the same degree as the WT enzyme (Table S2). This data strongly implicates the acetyl moiety binding site at the BC dimer interface as the site of overlap between acetyl-CoA and l-aspartate.

While acetyl-CoA primarily activates the BC domain half reaction, recent studies have implicated acetyl-CoA in modulating activity at the distantly located CT domain through subtle structural changes (8,9,38). For example, acetyl-CoA is known to exert a long range effect on the CT domain, reducing the apparent $K_{\mathrm{M}}$ of pyruvate in the presence of acetyl-CoA (9). In the current study, the CT domain inhibitors oxalate, 2-hydroxy-3-(quinoline-2-yl)propenoic acid, and phenylpyruvate all displayed a significantly reduced potency in the presence of acetyl-CoA, further reinforcing the long distance impact of acetyl-CoA binding on the CT domain active site. The increase in the IC₅₀ value for all three CT domain inhibitors suggests that acetyl-CoA decreases the binding affinity of CT domain ligands through a mechanism that may involve changes in carrier domain translocation (7). Interestingly, the acetyl moiety does not contribute to this long-range effect since CoA also increased the IC₅₀ value of CT domain inhibitors (Figure 6B). Consistent with this, the R21A and K46A SaPC mutations did not alter the ability of acetyl-CoA to diminish CT domain inhibition by oxalate (Figure 6C). Thus, the acetyl moiety binding site at the BC domain dimer interface does not directly contribute to this long-range communication with the CT domain active site. We suggest that the allosteric effect of the acetyl moiety binding site is localized to the BC domain half reaction (17).

We conclude that Arg₂₁, Lys₄₆, and Glu₄₁₈, three residues located in close proximity to the acetyl moiety binding site at the BC dimer interface, are important in binding acetyl-CoA, and are essential in mediating allosteric inhibition by l-aspartate but are not required for long range communication to the active site in the CT domain. Our results highlight the acetyl moiety binding site in SaPC as a critical allosteric “hot spot”. From a metabolic perspective, there is a well rationalized reason for focusing the competition between l-aspartate and acetyl-CoA at the acetyl moiety binding site. Feedback inhibition from l-aspartate serves to maintain homeostasis in the high energy “fed” state. Given that the acetyl-CoA/CoA ratio is particularly well suited to gauge the metabolic state of the cell (52), the most direct way for a feedback inhibitor to alleviate activation in the fed state is to compete at the acetyl moiety binding site that is directly responsive to the acetyl-CoA/CoA ratio. The current study provides a more complete and precise molecular description of allosteric regulation in PC, offering new possibilities for the design of selective inhibitors with the long-term potential to target bacterial infection, cancer, or type II diabetes.