The Dephospho-CoA Kinase of Mycobacterium tuberculosis: Molecular Mechanism and Function

Abstract

Tuberculosis (TB) continues to represent a significant global health danger, with the bacillus Mycobacterium tuberculosis (Mtb) identified as its causative agent. An essential metabolic helper, Coenzyme A (CoA) engages in multiple core cellular processes that are pivotal for both Mtb survival and its ability to cause disease. The terminal step of CoA biosynthesis is catalyzed by dephospho-CoA kinase CoaE, which phosphorylates dephospho-CoA to generate the active cofactor. However, the molecular mechanisms regulating Mtb CoaE catalytic activity remain poorly understood. Here, AlphaFold 3 was used to predict CoaE model. The Ramachandran plot and ProSA were used to validate the model. Subsequently, we performed molecular docking, followed by site-directed mutagenesis. We found that the S15A, R145A or N175A mutants showed markedly reduced activity, whereas the K14A mutation completely abolished activity. Additionally, deletion of residues 27–101 (Δ27–101) led to a significant loss of function. Our findings offer new insights into the structure and mechanism of Mtb CoaE, a promising target for designing selective antituberculosis agents.

1. Introduction

An intracellular microbe, Mycobacterium tuberculosis (Mtb), provokes tuberculosis (TB), which still ranks among the foremost lethal outcomes attributable to a solitary infectious agent worldwide. The disease is primarily attributed to members of the Mtb complex, a group of closely related mycobacterial species that collectively threaten human health. Extended therapeutic regimens for TB, compounded by the growing spread of extensively and multidrug-resistant variants, make it critically necessary to discover novel pharmacological targets and antibiotics with entirely distinct killing mechanisms. In this regard, essential metabolic pathways that are either unique to Mtb or structurally divergent from host counterparts have emerged as promising strategies for drug development.

An omnipresent and critically important cofactor, Coenzyme A (CoA) governs the transfer and activation of acyl groups within multiple central metabolic hubs, encompassing the tricarboxylic acid cycle, fatty acid formation and degradation, and the biosynthesis of the elaborate lipid moieties that make up the mycobacterial cell wall. , Mtb synthesizes CoA de novo from pantothenate (vitamin B5) through a canonical five-step pathway, in which the final, rate-limiting step is catalyzed by dephospho-CoA kinase (CoaE). , CoaE phosphorylates the 3′-hydroxyl group of dephospho-CoA (dCoA) using ATP to generate the active CoA molecule (Figure ). Given CoA′s essential role in Mtb viability and the substantial metabolic demand of maintaining the pathogen′s complex cell envelope, genetic or chemical disruption of this pathway presents a promising strategy for anti-tubercular therapy. − Although Mtb CoaE has been recognized as a high-confidence drug target, its structural and mechanistic features remain poorly known.

1.

Comparative alignment of dephospho-coenzyme A kinase sequences. This display relies on the canonical ClustalW chromatic rules, where invariant positions receive a deeper tint relative to more divergent sites. The reference proteins featured here are Q8E449, Streptococcus agalactiae serotype III (strain NEM316); P0DA45, Streptococcus pyogenes serotype M3 (strain SSI-1); Q8XIX0, Clostridium perfringens (strain 13/Type A); Q67R57, Symbiobacterium thermophilum (strain DSM 24528/JCM 14929 IAM 14863/T); Q8YW62, Nostoc sp. (strain PCC 7120/SAG 25.82/UTEX 2576); Q92BF2, Listeria innocua serovar 6a (strain ATCC BAA-680/CLIP 11262); P9WPA3, Mycobacterium tuberculosis (strain ATCC 25618/H37Rv); and Q5R0N3, Idiomarina loihiensis (strain ATCC BAA-735/DSM 15497/L2-TR).

Herein, the hydrodynamic diameter of CoaE was 6.1 ± 0.3 nm. CoaE model was generated using AlphaFold 3 and validated through Ramachandran plot analysis and ProSA assessment. According to molecular docking results, site-directed mutagenesis was performed. Mutants S15A, R145A or N175A exhibited markedly reduced enzymatic activity, whereas the K14A mutation completely abolished activity. Deletion of residues 27–101 (Δ27–101) also resulted in a substantial loss of function. Our results illuminate the molecular enzymology of Mtb CoaE and underscore the promise of leveraging its unique regulatory circuitry to develop next-generation anti-tubercular compounds.

2. Results

2.1. Dynamic Light Scattering Analysis of CoaE

A hydrodynamic diameter of 6.1 ± 0.3 nm (Figure ) was recorded for CoaE via dynamic light scattering (DLS) measurements on samples that had been centrifuged beforehand. This indicates to a monomeric state.

2.

Protein CoaE analysis using DLS. CoaE exhibits a hydrodynamic diameter of 6.1 ± 0.3 nm, indicating it exists in a monomeric state.

2.2. Predicting and Validating the CoaE Model

The CoaE model was predicted via AlphaFold 3 (Figure a). Unlike conventional homology modeling, AlphaFold 3 employs deep learning algorithms to achieve highly accurate and reliable protein structure predictions. To gauge the conformational reliability of the computationally derived CoaE structure, we turned to Ramachandran mapping (Figure b; Table ), which evaluates backbone φ/ψ angles for steric acceptability and corroborates native folding. The distribution census placed 95.5% of the residues in the premier allowed regions and 4.5% in the secondary permissive zones, with zero occurrences in either the generously tolerated or disallowed quadrants. This polypeptide also incorporated 1 terminal residue, 27 glycines, and 20 prolines (Figure b; Table ). As more than 90% of the amino acids sit within energetically optimal conformations, these results firmly attest to the model high trustworthiness and predictive accuracy.

3.

Predicting CoaE structure and evaluating model quality. The stereochemical and fold-quality of the CoaE model were examined through two complementary computational approaches. (a) The AlphaFold 3-generated structural architecture is illustrated in a ribbon representation from two viewpoints, with α-helices colored cyan, β-strands in magenta, and loop segments in pink. (b) Ramachandran mapping of backbone torsion angles then designated the core favored regions in red, while less preferred areas are shown in progressively fainter tints. Additionally, (c) ProSA validation yielded a global Z-score of −9.57 for the overall model.

1. Ramachandran Plots of the CoaE Model .

Residual properties	Number of residues	Total % of residues
Most favored region	343	95.5
Additional allowed region	16	4.5
Generously allowed region	0	0
Disallowed region	0	0

^aTwenty prolines are present, together with 27 glycines (rendered as triangles) and one terminal residue that does not include glycine or proline.

An additional assessment via the ProSA-Web service gave a Z-score of −9.57 (Figure c), signifying that the predicted fold is encompassed by the range of experimentally established structures and maintains a stable energetic profile. Collectively, these analyses confirm that the predicted CoaE model (Figure a) is suitable for subsequent functional and mechanistic studies.

2.3. The CoaE-ATP Complex Ligand-Binding Sites

A reliable model of CoaE was obtained through AlphaFold 3 (Figure a). This structural model subsequently formed the foundation for in silico docking trials carried out with AutoDock, , with the goal of estimating the CoaE-ATP complex. The docking results yielded a binding energy of −5.94 kcal/mol, indicating high confidence in the interaction. The predicted structure revealed that ATP (adenosine triphosphate) fits snugly within a positively charged cavity (Figure a). Several conserved residues (K14, S15, R145 and N175) within the binding pocket were identified as critical contributors to catalytic activity (Figure b).

4.

The binary complex model of CoaE-ATP. The substrate ATP (Adenosine triphosphate) is shown as sticks. (a) Colored by net charge, the surface depiction assigns red to negative, blue to positive, and white to neutral residues. (b) The ribbon representation includes a close-up of the binding sites, where dashed lines denote contacts between charged residues.

2.4. Enzymatic Characterization of CoaE

To assess the functional contribution of critical amino acid residues, we carried out targeted mutagenesis guided by the molecular docking outcomes. Mutants S15A, R145A or N175A exhibited markedly reduced enzymatic activity (Figure ), whereas the K14A mutation completely abolished activity (Figure ). These results highlight the critical role of positively charged residues, particularly K14 and R145, in catalysis. Their strategic positioning near the substrate stabilizes the negatively charged phosphate group of ATP and helps neutralize its charge. ATP interacts with the CoaE binding pocket primarily through the side chains of these residues. Collectively, these findings underscore the evolutionary conservation of positively charged residues within the dephospho-CoA kinase family.

5.

Catalytic activities of wild-type (WT) CoaE and its mutant variants. The activity of S15A, R145A, or N175A mutants was dramatically reduced relative to the wild-type (WT) protein. In contrast, the K14A mutation resulted in a complete loss of activity. The WT CoaE activity was set to 100%.

To elucidate the structural determinants critical for catalysis, we determined the kinetic parameters of the wild-type (WT) Mtb CoaE and its mutants (Table ). The WT CoaE exhibited a K _m of 55.64 μM and a K _cat of 2.86 min^–1. Notably, the K14A mutation resulted in a complete loss of detectable enzymatic activity, thus preventing the reliable determination of its kinetic constants. Among the active mutants, the S15A substitution showed a K _m of 59.84 μM and a K _cat of 0.68 min^–1, indicating that this residue primarily affects the catalytic turnover rather than substrate binding. In contrast, the R145A mutation substantially compromised both substrate affinity and catalysis, displaying an elevated K _m of 98.57 μM (approximately 1.7-fold higher than WT) and a drastically reduced K _cat of 0.29 min^–1, which corresponds to only ∼10% of the wild-type turnover number. Meanwhile, the N175A mutant retained near-wild-type substrate binding, with a K _m of 57.93 μM, while its K _cat was diminished to 0.72 min^–1, roughly 25% of the WT value. These suggest that R145 plays a pivotal role in both substrate recognition and the catalytic cycle, whereas S15 and N175 are predominantly involved in maintaining the optimal catalytic efficiency.

2. Kinetic Parameters of CoaE Constructs .

Proteins	K m (μM)	K cat (min–1)
WT (Wild-type)	55.64 ± 3.79	2.86 ± 0.14
K14A	ND	ND
S15A	59.84 ± 3.24	0.68 ± 0.05
R145A	98.57 ± 4.16	0.29 ± 0.03
N175A	57.93 ± 3.15	0.72 ± 0.03

^aThe kinetic parameters were determined by increasing the concentration of substrate ATP.

^bND: Not detectable.

2.5. ATP Is the Optimal Substrate among These Substrates

To evaluate the substrate specificity of CoaE, catalytic activity was measured using various nucleotide triphosphates, including ATP, TTP, GTP, CTP and UTP. The results demonstrated that CoaE exhibited the highest activity with ATP as the substrate (Figure ). These findings are consistent with previously reported observations. ,

6.

Substrate specificity of CoaE for phosphoryl donors. The activity is shown as a percentage of the activity toward ATP. The enzyme functional output is set at 100% under ATP-containing assay conditions.

2.6. Influence of Metal Ions on the Activity of CoaE

We investigated the influence of metal cations on the catalytic performance of CoaE (Figure ). Catalytic activity was markedly inhibited by Fe²⁺, Ca²⁺, and Li⁺, with reductions exceeding 80% (Figure ). Co²⁺, Mn²⁺, Ni²⁺, Zn²⁺, and Cu²⁺ also suppressed activity to varying extents, resulting in 40–80% inhibition (Figure ). In contrast, Na⁺ and K⁺ completely abolished CoaE activity (Figure ).

7.

Effects of different metal ions on CoaE activity. The activity of CoaE was inhibited to varying extents by different metal ions; however, Na⁺ or K⁺ caused a complete loss of activity. In the presence of Mg²⁺, the catalytic activity of the WT (wild-type) protein was standardized to 100%.

2.7. Deletion of Segment 27–101 (Δ27–101) Dramatically Reduces CoaE Activity

Segments 27–101 and 204–407 occupy the outer boundaries of the CoaE catalytic cleft (Figure a). When the 27–101 stretch was removed (Δ27–101), enzyme function was virtually abolished (Figure b), highlighting the indispensable contribution of this portion to overall activity. In contrast, deletion of residues 204–407 (Δ204–407) caused only a slight reduction in activity compared with the full-length enzyme (Figure b), suggesting that while this region is adjacent to the catalytic site, it plays a supportive rather than critical role.

8.

Effects of regions 27–101 and 204–407 on the catalytic activity of CoaE. (a) Residues 27–101 and 204–407 are highlighted with magenta and orange, respectively. (b) A severe reduction in activity was observed for Δ27–101, but only a small decrease for Δ204–407, with WT CoaE defined as 100%.

3. Discussion

In this work, the hydrodynamic diameter of CoaE was 6.1 ± 0.3 nm. Its model was predicted using AlphaFold 3 and validated through Ramachandran plot analysis and ProSA assessment. Guided by molecular docking results, site-directed mutagenesis was performed. Mutants S15A, R145A or N175A exhibited markedly reduced activity, whereas the K14A mutation completely abolished enzymatic function. Deletion of residues 27–101 (Δ27–101) also resulted in a substantial loss of activity. Taken together, these data afford a blueprint for the logical engineering of inhibitory molecules that target allosteric sites. Such inhibitors would specifically interrupt mycobacterial CoA synthesis while leaving the host own metabolic network untouched.

We chose molecular docking on the AlphaFold 3-predicted structure over a simple structural superposition onto ATP-bound homologues (PDB: 1UF9, 8U96) because AlphaFold 3 generates a species-specific model that captures unique binding-pocket conformations and avoids template-derived biases from interspecies differences, especially in loop and side-chain orientations critical for ligand recognition, and because recent benchmarks confirm that AlphaFold 3 achieves near-experimental accuracy for pocket architectures when confidence scores are high, making it a reliable docking platform. To validate our results, we superposed our AlphaFold 3 model onto the available homologous crystal structures and found that the ATP-binding pocket, including the loop and key catalytic residues, is highly conserved; the docking-derived ATP pose places the phosphate groups near the loop backbone amides and the adenine ring in a hydrophobic cleft, closely matching the experimentally observed ATP orientations in 1UF9 and 8U96, with key hydrogen-bond and electrostatic interactions also reproduced, supporting the biological plausibility of our model. We acknowledge that AlphaFold 3 has limitations in small-molecule positioning, but we mitigated this by restricting docking to high-confidence regions (pLDDT > 90), performing extensive conformational sampling with pose clustering, and cross-validating the final binding mode against the experimental homologues. This integrated approach, combining the predictive power of AlphaFold 3 with experimental anchoring-provides a robust framework for identifying the CoaE-ATP interaction, and the strong consistency between our predicted binding mode and the experimentally determined homologous structures reinforces our assignment and enables meaningful comparison of conserved catalytic residues across the dephospho-CoA kinase family.

The biosynthesis of coenzyme A (CoA) is a vital metabolic pathway critical for Mtb survival. The structural and regulatory distinctions of mycobacterial dephospho-CoA kinase, compared to its human homolog, make it a highly attractive therapeutic target. We applied an integrated strategy combining biophysical analyses, computational structural modeling, and site-directed mutagenesis to elucidate the molecular determinants underlying CoaE catalytic activity. Our results provide new mechanistic insights into the enzymology of this essential enzyme.

Site-directed mutagenesis guided by molecular docking identified several residues essential for CoaE enzymatic activity. The K14A mutation completely abolished activity, highlighting the indispensable role of Lys14 in catalysis. This finding is consistent with UniProt annotations, which designate residues 11–16 as the canonical ATP-binding P-loop motif. Within this motif, Lys14 is predicted to coordinate the negatively charged phosphate groups of ATP, and its substitution likely disrupts nucleotide binding, thereby fully blocking the phosphotransfer reaction. Mutations S15A, R145A or N175A caused substantial reductions in activity, further delineating the functional organization of the active site. Ser15, positioned adjacent to the catalytic lysine, is likely involved in orienting ATP or stabilizing the transition state. Similarly, Arg145 appears critical for anchoring the nucleotide substrate, reflecting functional conservation among homologous residues. The pronounced loss of activity in the N175A mutant indicates that Asn175 contributes to proper positioning of dephospho-CoA or forms part of a hydrogen-bonding network within the catalytic pocket, both of which are essential for efficient phosphoryl transfer.

A key mechanistic insight from our study is the dramatic loss of catalytic activity upon deletion of residues 27–101 (Δ27–101) within the N-terminal domain. Although this region is located outside the canonical loop and core catalytic motifs defined by sequence homology, its removal severely compromises enzymatic function. These results provide strong experimental evidence that the N-terminal segment serves a role beyond acting as a structural linker. ,, The deletion likely disrupts interdomain communication, rendering the enzyme conformationally inactive, analogous to the inactive trimer, even with substrate. This is in line with the unique evolutionary adaptation of mycobacterial CoaE, which possesses an extended N-terminal segment absent in simpler bacterial orthologs, likely facilitating regulatory interactions with the downstream UPF0157 domain. ,

Our structural and biochemical findings corroborate and substantially expand upon previous studies of Mtb CoaE. (1) The structural model predicted by AlphaFold 3 not only validates the previously constructed homology model, but also provides a significantly improved structural framework. This is evidenced by superior Ramachandran plot statistics (94.2% of residues in the most favored regions) and a more favorable ProSA Z-score (−9.57). This enables the successful docking of ATP into the loop region for the first time. This interaction had previously been elusive due to the enzyme requirement for prior DCoA saturation. (2) The extended multiple sequence alignment across eight bacterial species corroborates the strict conservation of the Walker A motif and key residues, including K14, D32, and R140, while also revealing additional conserved regions within the CoA and lid domains that warrant further investigation. Notably, this work confirms the essential role of K14 in catalysis. However, the complete loss of activity observed for K14A suggests that differences in experimental conditions or protein preparation may exist. (3) A systematic evaluation of metal ion specificity shows that Mg²⁺ is the optimal cofactor. Mn²⁺ supports only partial activity, while Na⁺ and K⁺ completely inhibit catalysis. These findings substantially extend the initial observation of Mg²⁺ dependence and establish a clear hierarchy of metal ion preference. (4) The domain deletion analysis, which revealed that Δ27–101 dramatically reduces activity and that Δ204–407 only moderately impairs catalysis, is consistent with the paradigm that the C-terminal UPF0157 domain plays an important role in proper protein folding and soluble expression, rather than in catalysis itself. This is further corroborated by the observation that the NTD-GST fusion protein retained 95% of wild-type activity. (5) The phosphate donor specificity profile, with dATP supporting approximately 65% of ATP-driven activity and CTP, GTP, and UTP exhibiting minimal activity (<10%), fully recapitulates the findings of Walia et al., reinforcing the conclusion that CoaE displays stringent specificity for adenine-containing nucleotides. Together, these results strengthen the existing mechanistic framework, provide new structural insights, and expand the range of functionally characterized residues. This offers a more complete foundation for developing structure-guided antimycobacterial drugs that target this essential enzyme.

The collective findings reported here have important implications for the design of novel antitubercular agents. The essential role of Lys14 in catalysis highlights the ATP-binding pocket as a well-established target for competitive inhibition. However, the distinctive vulnerability of CoaE resides in its intricate regulatory architecture. Our observation that deletion of residues 27–101 mimics the inhibitory effect of CTPwhich prevents the monomer-to-trimer transitionstrongly supports the strategy of targeting the allosteric network rather than the highly conserved catalytic site. , The residues 27–101 constitute a functionally critical region that may contribute to the structural dynamics and/or allosteric modulation of CoaE rather than serving solely as a passive structural scaffold. Nevertheless, we acknowledge that deletion analysis is a crude tool. A clear dissection of the deletion mutant’s architectural versus modulatory functions will eventually demand high-resolution biophysical methods, for example, crystallographic analysis or solution NMR. Additionally, targeted point mutations within this region are necessary to fine-map the functional determinants. These experiments are currently underway in our laboratory.

In summary, this study enhances our understanding of the structure–function relationship of Mtb CoaE by identifying essential catalytic residues and experimentally characterizing a critical N-terminal regulatory module. These findings illuminate fundamental aspects of mycobacterial CoA metabolism and further validate Mtb CoaE as a high-priority drug target, offering a framework for the rational development of allosteric inhibitors that leverage its unique regulatory mechanisms to combat drug-resistant tuberculosis.

4. Conclusion

In this study, the solution-phase hydrodynamic diameter of CoaE was 6.1 ± 0.3 nm. The structural model generated by AlphaFold 3, validated through Ramachandran plot and ProSA analyses. Site-directed mutagenesis revealed that Lys14 is essential for catalysis, while mutations S15A, R145A or N175A caused substantial reductions in activity. Notably, deletion of residues 27–101 (Δ27–101) dramatically compromised function. The structural and mechanistic characterization of Mtb CoaE identifies this enzyme as a promising target for designing selective anti-tuberculosis drugs.

5. Materials and Methods

5.1. Analysis of Bioinformatics

We performed multiple sequence alignments on the CoaE amino acid sequence (UniProt P9WPA3; Figures S1, S2) using ClustalW with its standard parameters, and then rendered the resulting alignments with ESPript 3.0.

5.2. Prediction and Assessment of the CoaE Structure

AlphaFold 3 was used to predict the 3D structure of CoaE (Figure S3) from its protein sequence (UniProt P9WPA3). The resulting molecular images were created and viewed through PyMOL v2.3.4 (https://www.pymol.org/2/). Geometric validation consisted of a Ramachandran assessment employing PDBsum, which examines backbone dihedral pairs (φ/ψ) to assess stereochemical fitness, this revealed residues in energetically stable conformations as well as those in less ideal positions, thus giving a numerical measure of the model reliability. Additionally, ProSA (Protein Structure Analysis) was used to further assess model quality; this widely adopted tool evaluates predicted protein structures based on statistical analyses derived from experimentally determined X-ray and NMR structures.

5.3. Preparation of Recombinant CoaE

The cloning, expression, and purification of CoaE (UniProt accession P9WPA3) were performed following previously described protocols with minor modifications. , We inserted the amplified gene into the pET28a vector at its NdeI and HindIII sites, then transformed the construct into E. coli BL21­(DE3) competent cells for heterologous expression. T7-promoter and terminator primers were used in sequencing to verify the plasmid integrity. For optimal production of the His₆-tagged protein, we grew cultures at 37 °C until the optical density at 600 nm reached 0.6–0.8, then induced with 0.1 mM IPTG (Isopropyl-beta-D-thiogalactopyranoside) and incubated for 8 h at 18 °C. Cells were pelleted, resuspended in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 10% glycerol, pH 7.4), and disrupted by sonication at 4 °C over two 10 min cycles. After clearing the lysate by centrifugation, we loaded it onto Novagen Ni-NTA resin and eluted the captured protein with 200 mM imidazole. The eluate was then dialyzed against a storage buffer (20 mM Tris, 150 mM NaCl, 5% glycerol, pH 7.4) and concentrated for subsequent use.

5.4. DLS (Dynamic Light Scattering) Characterization

We assessed the oligomerization status of CoaE by measuring its hydrodynamic size via dynamic light scattering (DLS) on a Malvern Zetasizer, following a previously described protocol with minor adjustments. The sample was first brought to ∼2.3 mg/mL in Tris-HCl buffer (20 mM Tris-HCl, 150 mM NaCl, pH 7.4) and cleared of aggregates by a 5 min spin at 18,000 rpm and 4 °C. The clarified material was transferred to a 1 cm quartz cuvette, and scattering data were collected across 30 consecutive acquisitions after a 120-s thermal equilibration. Raw DLS traces were processed with Zetasizer software (v6.20), which generated regularized distribution profiles, with the diameter value being tracked at each acquisition step.

5.5. Molecular Docking

Molecular docking of ATP (adenosine triphosphate) within the CoaE binding pocket was performed (Figure S4) using AutoDockTools 1.5.7 and AutoDock 4.2.6. ,, The Genetic Algorithm (GA) was configured for 1,000 runs, a population size of 200, generations of 30000, and energy evaluations of 3000000. We next examined the molecular docking outputs and rendered the CoaE-ATP contacts using PyMOL v2.3.4 (https://www.pymol.org/2/). The calculated binding energy of −5.94 kcal/mol lends credence to the docking model’s validity. This energetic value, given in kcal/mol, quantifies the ligand-target affinity; more negative numbers correspond to tighter associations and enhanced complex stability. , As a critical indicator, it helps evaluate functional modulation and directs the selection of candidate bioactive molecules for downstream investigations.

5.6. Enzymatic Activity Assay

We measured the catalytic function of CoaE and its mutant forms using a refined version of established procedures. , All reaction mixtures were incubated at 25 °C, with one unit of activity corresponding to 1 μmol of product generated per minute; parallel blanks lacking dephospho-CoA (DCoA) were subtracted as baselines. Two distinct formats were applied to evaluate CoaE activity: the conventional double-coupled pyruvate kinase/lactate dehydrogenase (PK/LDH) system and a single-coupled α-ketoglutarate dehydrogenase (α-KDH) assay. The latter follows CoA production by recording the absorbance rise at 340 nm due to NADP⁺ reduction. Each α-KDH reaction vessel contained ATP (10 mM), DCoA (0.5 mM), MgCl₂ (10 mM), KCl (20 mM), α-ketoglutarate dehydrogenase (2 U), α-ketoglutarate (2 mM), NADP⁺ (0.3 mM), and thiamine pyrophosphate (0.3 mM) in 50 mM Tris-HCl (pH 8.0). We further corroborated enzyme function by analyzing product formation in a separate reaction mix composed of DCoA (0.25 mM), ATP (0.25 mM), γ³²P-ATP (0.5 μCi), MgCl₂ (10 mM), and DTT (1.5 mM) in the same Tris buffer, then resolving the products via descending paper chromatography on Whatman 3 MM paper with a mobile phase of isobutyric acid: 0.5 N NH₄OH (100:60) plus 1 mM EDTA.

5.7. Enzymatic Activity Assay of CoaE Site-Directed Mutants

Primers targeting CoaE for mutagenesis (Table S1) were generated for PCR and obtained commercially from Sangon Biotech (Shanghai, China). The pET28a plasmid harboring the CoaE gene was used as the template, and mutagenesis was performed using Q5 DNA polymerase. We reinserted the mutagenized CoaE genes into the pET28a vector and then expressed and purified the proteins using the identical protocols employed for the WT counterpart (Figures S5, S6). The presence of the intended nucleotide changes was verified by DNA sequencing performed at Shanghai Sangon Biotechnology (China). Enzymatic activity of the mutant proteins was assayed under the same conditions as those applied to wild-type CoaE.

5.8. Accession Codes

The UniProt identifier for the CoaE protein of Mycobacterium tuberculosis (strain ATCC 25618/H37Rv) is recorded as P9WPA3.

5.9. Statistical Analysis

All experiments were performed in triplicate or more. Data are presented as mean ± standard deviation (SD). Statistical analyses were conducted using Origin 8.5, Microsoft Excel 2013, and SPSS 19.0. One-way ANOVA followed by Bartlett’s test was employed to assess statistical significance. Differences with p < 0.05 were considered statistically significant, while p < 0.01 indicated high statistical significance.

The data presented in this study can be found in the article itself and the Supplementary Materials.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c05535.

• Figure S1: The amino acid sequence of CoaE (UniProt code P9WPA3); Figure S2: The optimized sequence of gene CoaE; Figure S3: The per-residue Local Distance Difference Test (pLDDT) was used to evaluate the confidence of the CoaE structure predicted via AlphaFold 3; Figure S4: Structural alignment; Figure S5: Expression and purification of Δ204–407; Figure S6: Expression and purification of Δ27–101; Table S1: Primers used for generating site-directed mutants of CoaE (PDF)

Conceptualization, D.L.; Data curation, D.L.; Formal analysis, D.L.; Investigation, D.L., H.D., H.S.; Methodology, D.L.; Resources, D.L.; Software, D.L.; Validation, D.L.; Writingoriginal draft, D.L.; Writingreview and editing, D.L. All authors have read and approved the final version of the manuscript.

Our research is financially supported by grants from the third batch of the Tianchi Talent Young Doctoral Research Grant, Xinjiang Autonomous Region (2025QNBS001), and the Talent Team Support Initiative for Scientific Research and Innovation Platforms of the Xinjiang Talent Development Fund (XJRCXYCTD2025-2).

The authors declare no competing financial interest.