Comparison of Magnesium and Manganese Ions on the Structural and Catalytic Properties of Human DNA Polymerase Gamma

Abstract

DNA polymerases are essential enzymes responsible for accurate genome replication and repair, with divalent metal cofactors playing a crucial role in their catalytic function. Polymerase γ (Pol γ) is the primary DNA polymerase in mitochondria, ensuring the faithful replication of mitochondrial DNA. The choice of metal cofactor, typically magnesium (Mg²⁺) or manganese (Mn²⁺), influences its structural stability, enzymatic activity, and fidelity. In this study, we employed molecular dynamics (MD) simulations and hybrid quantum mechanics/molecular mechanics (QM/MM) calculations to investigate how Mg²⁺ and Mn²⁺ affect the flexibility, active site stabilization, and catalytic efficiency of Pol γ. Intermolecular interaction analysis of individual residues is consistent with experimental mutagenesis reports and highlights the importance of specific residues, many of which are evolutionarily conserved, and some are involved in pathogenic mutations. It is also observed that Mn²⁺ enhances catalytic efficiency, exhibiting higher exoergicity (−3.65 kcal mol⁻¹ vs −1.61 kcal mol⁻¹ for Mg²⁺) and a lower activation barrier. Intermolecular interaction analysis reveals that Mn²⁺ provides larger stabilization of the transition state and product complex, favoring reaction progression. Investigation of the effects of the electric field in the active site suggests that the O3′ atom on the DNA primer base experiences larger polarization in the system with Mn²⁺ ions when compared to Mg²⁺, with dipole directions consistent with the catalytic reaction progress. Our findings highlight a trade-off between structural stability and catalytic efficiency, providing insights into the role of metal ions in mitochondrial polymerase function and their implications for mutagenesis and mitochondrial disorders.

Graphical Abstract

1. INTRODUCTION

Accurate maintenance and transfer of genetic information are crucial to the maintenance of genomic stability. Proper replication of DNA is important for progenitor cells to have the same genetic information. Enzymes that facilitate proper replication as well as repair of DNA are called DNA polymerases. These biological catalysts synthesize new DNA strands by adding nucleotides to the preexisting primer in a sequence complementary to the template strand with high fidelity.¹ Polymerases are usually classified based on their structural features as well as functions into seven different families: A, B, C, D, X, Y, and RT.

The general reaction mechanism catalyzed by DNA polymerases includes an elementary step involving the nucleophilic attack of the O3′ atom of the terminal primer nucleotide and Pα of the incoming nucleotide. This leads to the formation of an O–P bond and pyrophosphate (PPi), followed by PPi unbinding and translocation of the DNA to make space for the next nucleotide to come in, enabling further DNA elongation. The active sites of most DNA polymerases, where this reaction occurs, are generally similar. They usually consist of the terminal primer nucleotide, the incoming nucleotide, conserved carboxylates (Asp and Glu), and metallic cofactors. In most DNA polymerases, it is seen that this reaction is facilitated by two metallic cofactors.^2–6

The two metals present in the active site serve specific purposes.⁷ The ions bound to the so-called A- and B-sites (Figure 1) play crucial roles, including polarizing the 3′ OH, resulting in lowering of its pK_a, which facilitates its deprotonation and subsequent nucleophilic attack on the α-phosphate of the incoming nucleotide. The metals are also responsible for neutralizing the developing negative charge brought about by the nucleotidyl transfer reaction. This B-site metal is coordinated to the α-, β-, and γ-nonbridging phosphate oxygens of the incoming nucleotide and helps in the removal of PP_i after the reaction has occurred.

Figure 1.

3D structure of human DNA pol γA with Mg²⁺ ions in the active site, modeled after the 4ZTZ crystal structure. The active site is magnified and catalytically significant residues in the active site are shown.

Physiologically, Mg²⁺ ions are the most probable cofactors due to their higher concentration in cells, but sometimes other ions can also bind to the active site. Different ions can affect DNA pols in different ways, such as changing the ground-state binding affinity of the incoming nucleotide,⁸ decreasing base selectivity by promoting misincorporation,⁹ as well as decreasing base excision rate,¹⁰ among others. Various studies on different polymerases incorporating different ions have shown reduced DNA replication fidelity, as well as mutagenic and carcinogenic properties of ions other than Mg²⁺.^9,11–21

Several studies have shown that the incorporation of Mn²⁺ ions in place of Mg²⁺ enhances the rate of the reaction.^19,22 A computational study of the reaction mechanism of DNA pol λ by Cisneros et al.²³ with both Mg²⁺ and Mn²⁺ ions showed that the reaction barrier of the system catalyzed with Mn²⁺ ions was lower than that of the system with Mg²⁺ ions, consistent with experimental findings. Experimental investigation on DNA pol γ, the mitochondrial DNA polymerase, has also shown an increased rate of polymerization in the presence of Mn²⁺ ions when compared to Mg²⁺ ions (Park, Torres-Baruch, and Yin, personal communication). Hence, investigating the differences that bring about this change can provide insights into the effects of these ions.

Mitochondrial DNA replication is extremely important, as the mitochondrial genome encodes essential components for oxidative phosphorylation and energy production. The replication of mitochondrial DNA is mediated by polymerase γ (pol γ).^24–26 Pol γ is classified under the A family of polymerases and possesses both replicative and editing functionality, providing high-fidelity replication and repair of DNA. This is crucial, as errors can lead to mutations associated with various diseases, including neurodegenerative disorders and other metabolic syndromes, as well as loss of catalytic function.^27–30

Pol γ is a heterodimer with a catalytic subunit (pol γA) and a homodimeric regulatory subunit (pol γB). It shares structural elements with other replicative DNA polymerases, such as the thumb, finger, and palm domains^25,31 (Figure 1). The active site, where the addition of a new incoming nucleotide occurs, consists of the terminal primer, conserved carboxylates (Asp⁸⁹⁰ and Asp¹¹³⁵), the incoming nucleotide, as well as two cofactors, which are generally divalent metal ions. These divalent metal ions (generally Mg²⁺ or Mn²⁺) are essential for primer extension as well as for exonuclease activity.^15,32–34

In this study, we aim to investigate the effects of the two ions on the dynamics and kinetics of the system using computational simulations based on classical molecular dynamics (MD)^{14,30,35–42} and hybrid QM/MM calculations.^{13,23,43–48} The sections below outline the approach we used for MD and QM/MM, elaborating on the various analyses used. Further sections discuss the results for each system, with a relative perspective on the two metals and their contributions. Finally, we end with some remarks that summarize our findings.

2. COMPUTATIONAL METHODS

2.1. Structural Preparation and Molecular Dynamics Simulations.

This work is an extension of a previous work by Park et al. that combined experimental and computational analyses, where we investigated the effects of a single point mutation resulting in DNA pol γ R853A.³⁰ In that work, the structural preparation, including gap filling, and ionic concentrations were described in detail.³⁰ The system preparation and resulting ensembles from the molecular dynamics simulations for the system with Mg²⁺ are the same as the ones used in that study. Briefly, for the system preparation, DNA Polymerase γ (pol γ) was modeled based on the 4ZTZ crystal structure.³¹ Only the catalytic subunit, pol γA, was modeled and used for further calculations. Experimental results have shown that pol γA is catalytically competent for single nucleotide addition, and thus the B subunit is not necessary for catalysis.⁴⁹ The missing regions were modeled using Rosetta Fold.⁵⁰ Multiple structures were created, and the best structure based on preliminary visual inspection, followed by determination of the RMSD after overlay with the crystal structure, was selected as described in our previous work.³⁰ The two ions in the active site were then introduced, and two different ternary systems were generated: one with Mg²⁺ ions in the active site and the other with Mn²⁺ ions.

The parameters for the incoming nucleotide were previously reported.¹⁴ Protonation states for ionizable residues were calculated with ProPKA,⁵¹ and hydrogens were added based on the protonation states from the previous step, as well as clashes were checked and resolved using MolProbity.⁵² The LeAP⁵³ module of AMBER18 was used to neutralize and solvate the system. The ff14SB,⁵⁴ OL15,⁵⁵ and TIP3P⁵⁶ force fields were used for protein, DNA, and water (along with ions), respectively. Based on the volume of both systems, 73 Mg²⁺ and Mn²⁺ ions and 146 Cl⁻ ions were added to the systems to make 20 mM solutions of MgCl₂ and MnCl₂, respectively. This concentration was used for DNA elongation assays in our previous work, and we have maintained it for consistency.³⁰ The final atom count was 308356 atoms for the system with Mg²⁺ ions and 310729 atoms for the Mn²⁺ system.

The MD simulations for both systems were carried out using AMBER18’s pmemd.cuda.⁵⁶ Minimizations were carried out for a total of 10,000 cycles, with 5000 cycles using the steepest descent algorithm and the other 5000 cycles using the conjugate gradient. This was followed by heating the system slowly to 300 K using Langevin dynamics,⁵⁷ with a collision frequency of 2 ps⁻¹. All the atoms were restrained with a force of 100 kcal mol⁻¹ Å–2 during this time. These restraints were gradually reduced (see Table S1). The MD productions for each of the systems were run in triplicate. Each replicate was simulated for 500 ns, saving 100 frames per nanosecond, giving a total of 50,000 structures for each replicate and 150,000 structures for each system.

The residues in the active site showed some distortion after the removal of the restraints. Hence, these residues, namely the incoming nucleotide, Asp⁸⁹⁰, Asp¹¹³⁵, the two metal ions, and the 3′-OH terminus, were restrained with an additional 0.25 kcal mol⁻¹ Å–2 for an additional 25 ns. Once the system was stabilized, the restraints were removed, and both systems were subjected to 500 ns NPT MD in triplicate with a 2 fs time step, resulting in a total of 1.5 μs. All bonds involving hydrogen were treated with SHAKE.⁵⁸ Long-range Coulomb interactions were handled with the smooth particle mesh Ewald (PME)⁵⁹ method, and long-range van der Waals interactions were approximated using the default isotropic correction⁶⁰ (in AMBER), with a 10 Å cutoff for nonbonded interactions.

The MD simulations were analyzed using the CPPTRAJ⁶¹ suite of AMBER. These analyses included RMSD, RMSF, dynamic cross-correlation, and NMA. Various Python and R libraries were used to organize and analyze the raw data, as well as tools such as Gnuplot⁶² and Matplotlib⁶³ were used to visualize the data. AMBER-EDA (Energy Decomposition Analysis) was carried out to determine the stabilizing/destabilizing effects of residues on the active site based on nonbonded (Coulomb and van der Waals) interactions using Amber-EDA.⁶⁴ K-means clustering analysis was carried out on all 150,000 structures of the two systems, with the O3′–Pα distance and the O3′–Pα–O3α angle as the two dimensions. Distances less than 3.2 Å and angles greater than 160° were considered to be catalytically viable, and 5 representatives were chosen from each of the systems for further QM/MM calculations. Multisequence alignment (MSA) was performed using ConSurf.⁶⁵

2.2. QM/MM Calculations.

All QM/MM calculations were carried out using LICHEM^66,67 to interface Gaussian16⁶⁸ and TINKER.⁶⁹ The MM region of both systems was represented with the ff14SB force field, and the active MM region was 30 Å from the O3′ of the terminal primer. The remaining atoms were “frozen”, meaning they were not optimized. The QM region of the Mg²⁺ system, consisting of the two metals, incoming nucleotide, terminal 3′, conserved carboxylates, and coordinating waters of both metals and triphosphates, was treated at the ωB97X-D/6–31G(d,p) level. For the Mn²⁺ system, the QM region (similar atom selections to the Mg²⁺ system) was treated at the ωB97X-D level with LANL2DZ for the Mn²⁺ ions and 6–31G(d,p) for all other atoms, for a preliminary optimization for the sake of efficiency. Once converged, further optimizations were performed with ωB97X-D/6–31G(d,p) for all QM atoms, including the Mn²⁺ ions. The metals in the active site for the Mn²⁺ system were modeled in a high-spin state (ferromagnetic, multiplicity of 11), based on previous studies of similar systems.^23,70–72 The long-range electrostatic correction (LREC)⁷³ with a 12 Å cutoff was used for the QM calculations, whereas PME was used for the MM calculations. The pseudobond approach was used in all places where covalent bonds between the QM and MM subsystems were cut,^74–76 and interactions between the QM and MM regions were treated with electrostatic embedding. The Davidson–Fletcher–Powell (DFP) algorithm was used for all single-structure geometry optimizations. The QM convergence criteria were: an RMS deviation of 0.001 Å, an RMS force of 0.005 hartree/Bohr, and a maximum force of 0.015 hartree/Bohr. The MM RMS deviation was 0.1 Å. The number of QM atoms for the Mg²⁺ system is 174, with 3536 active MM atoms, whereas for the Mn²⁺ system, the number of QM atoms is 171, with 3579 active MM atoms based on the number of waters observed to be hydrogen-bonded to the incoming nucleotide (Figure S1).

An iterative optimization protocol was implemented in LICHEM on the QM region and all atoms within 30 Å of the O3′ on the five chosen representatives of each system,^66,67 and the representative with the lowest energy was selected as the starting structure for product formation. This method of product generation has been validated in previous studies.^{13,23,43,44,77} The product structures were generated from the reactant structures using GaussView and optimized using the same iterative optimization approach.

The optimized end points were then used to determine the minimum energy path using the quadratic string method (QSM).⁷⁸ The initial guess for the intermediate structures for QSM was obtained using a linear interpolation between the end points. The QM region is optimized first, and then the MM region is optimized using the restrained-MM path optimization protocol implemented in LICHEM.⁶⁷ The criteria for the active MM region remained the same as those of the single structures, 30 Å from the O3′ of the terminal primer. The convergence criteria remained the same as those for the QM/MM optimization of the representative structures. The remaining atoms were not considered in the optimization. The initial restraint was set to 50 kcal mol⁻¹ Å⁻² on the MM region and relaxed gradually. The restraint is essential as the initial path is obtained using linear interpolation, with the optimized reactant and product structures as the end points.⁶⁷ For this, the MM region for all intermediate structures remains the same as that of the reactant rather than the product. If there is a significant difference between the reactant and product MM regions, then there might be a disconnect in the MM region during path optimization. Hence, restraints are added and slowly reduced (by half every QSM cycle), until 2 kcal mol⁻¹ Å⁻², after which all restraints are removed and at least two more optimization cycles occur, which helps in obtaining a smooth reaction path.⁶⁷ Including the end points (reactant and product), the reaction path had 16 structures in total, with 14 intermediates connecting the reactant and the product.

2.3. NCI and ELF Analyses.

NCI (noncovalent interactions) and ELF (electron localization function) analyses were carried out on all 16 beads of the reaction path for a deeper understanding of the electronic structure of the systems along the reaction path.⁷⁹ NCI analysis was performed with the promolecular density method,⁷⁵ as implemented in the Multiwvfn V 3.8 program.⁸⁰ This analysis provides qualitative data on chemical bonds and weak noncovalent interactions. The NCI surfaces are usually depicted on an RGB scale, with green and blue surfaces indicating weak and strong attractive interactions, whereas the red color indicates repulsive interactions. For ELF, the wave function was obtained for all 16 optimized beads, and then basin analysis implemented in Multiwvfn V 3.8 was used. Basins associated with catalytically relevant atoms in the reactant, approximate transition state, and product were further analyzed by obtaining the sum of dipoles in the specified basins, using an approach reported in a previous study.⁴³ Information regarding individual basins is provided in Tables S2 and S3. The images were rendered by using VMD.

2.4. Energy Decomposition Analysis (EDA).

EDA was performed with AMBER-EDA⁶⁴ and analyzed using R. For the MD simulations, the noncovalent interactions between selected residues and the rest of the protein were calculated as an average over the complete ensemble for each system. For the QM/MM-calculated structures, EDA was also performed on the reactant, product, and approximate TS obtained from the QSM-optimized paths for both systems to identify the effects of individual residues on the reaction path. A short 10 ns MD was run on all three structures after obtaining the respective parameters using PyRED.

3. RESULTS AND DISCUSSIONS

3.1. Molecular Dynamics Simulations.

RMSD and RMSF calculations were performed with the completed crystal structure as the reference. Since the replicates showed similar behavior, replicate 1 was chosen as a representative replicate for RMSD. The RMSF plot was obtained using the difference between the averaged RMSF of both systems. Per replicate structural analysis is presented in the Supporting Information (Figures S2–S7) along with averaged RMSD, RMSF, and metal-coordinating distances in the active site with error bars (Figures S8–S14). Small error bars in all analyses suggest consistency between replicates and validate the choice of a representative replicate.

The RMSD of the backbone for both systems for the polymerase domains (thumb, fingers, and palm) reveals the stability of the systems over the simulations (Figure 2B,C). The thumb domain shows higher RMSD in both systems. Of the two regions that comprise the thumb domain (441–476 and 786–816), regions 441–476 are responsible for the higher deviation. This region is present in the domain that is responsible for interacting with the pol γB subunit. Hence, the lack of pol γB is likely the reason for a higher RMSD in that region. However, when investigating per residue fluctuations of both systems, it was noticed that the system with Mn²⁺ shows higher flexibility in most of the protein, other than regions in the spacer and exo domains (Figure 2D,E). These regions are seen to have Mn²⁺ ions around them for most of the simulation time, which helps explain the reduced fluctuations compared with those of the Mg²⁺ system.

Figure 2.

Pol γ structure with color-coded domains corresponding to the RMSD traces (A). RMSD of the backbone of the thumb, fingers, palm, and DNA of the system with Mg²⁺ ions (B) and the system with Mn²⁺ ions (C). (D) Difference RMSF plot showing per residue fluctuation change between the system with Mn²⁺ ions and Mg ions. (E) Normalized values of the difference RMSF shown as a heatmap on the structure. Panels (A) and (B) of this figure have been reprinted, with permissions, from Park et al.,³⁰ copyright (2024), AAAS.

The dynamic cross-correlation analysis suggests specific differences in the dynamics of the two systems (Figure 3). It is seen that the exo domain shows higher correlated motion with the fingers domain and the palm domain in the system with Mn²⁺ ions (encircled in black), whereas the N-terminus domain shows higher correlation with the fingers domain in the system with Mg²⁺ ions (encircled in purple). The fingers and thumb domains show higher anticorrelated movement with respect to each other in the system with Mn²⁺, whereas the palm domain shows higher anticorrelation with fingers in the system with Mg²⁺ ions (encircled in green). Normal Mode Analysis shows that the mode with the highest contribution to the motion represents a breathing motion, while the second highest contribution shows a rocking motion (Video in Supporting Information).

Figure 3.

Difference cross-correlation analysis between replicate 1 of Mg²⁺ and Mn²⁺ ions (Mn−Mg), showing correlated movement (left) and anticorrelated movement (right). Regions in red showing higher correlation/anticorrelation in the system with Mg²⁺ ions and regions in blue show higher correlation/anticorrelation in the system with Mn²⁺ ions.

Structural analysis of specific fragments in the active site was carried out to determine the prevalence of catalytically competent structures sampled by both systems. K-means clustering across all three replicates of both systems, with a focus on population distribution relative to catalytic distances (O3′–Pα) and angles (O3′–Pα–O3α) was carried out to investigate catalytic propensity. It is seen that a considerable portion (~39% in the system with Mg²⁺ and ~31% in Mn²⁺) of the calculated ensembles shows catalytic competency, as defined by a catalytic distance/angle of O3′–Pα < 3.2 Å and O3′–Pα–O3α > 160°. Interestingly, a higher percentage of catalytically competent structures is observed for the Mg²⁺ systems compared with Mn²⁺ (Figure 4). Additionally, it is seen that the system with Mn²⁺ ions samples a wider distribution of both the catalytic distance and angle during the simulation. Overall, 58,469 catalytically competent structures were obtained for the Mg²⁺ system, and 46,536 catalytically competent structures were obtained for the Mn²⁺ system, from a total of 150,000 structures for each.

Figure 4.

K-means clustering analysis based on the O3′–Pα and O3′–Pα–O3α angles of the simulations containing Mg²⁺ (left) and Mn²⁺ (right) ions. Populations of each cluster are shown in the top-right.

Energy decomposition analysis (EDA) can reveal residues that have favorable and unfavorable interactions with the active site of both systems. A difference of the sum of the total nonbonded intermolecular interactions $\left(\Delta E_{\text{Mn}-\text{Mg}}=E_{\text{Mn}(\text{II})}^{\text{non}-\text{bonded}}-E_{\text{Mg}(\text{II})}^{\text{non}-\text{bonded}}=53.7{\text{kcal}/\text{mol}}^{-1}\right)$ with respect to the incoming nucleotide, the terminal primer, and the two metal cofactors shows increased favorable interactions of the protein’s environment on the system with Mg²⁺ when compared to Mn²⁺; however, this effect is reversed for the EDA analysis using the results from the QM/MM simulations (see below). Residues having a significant contribution (ΔE = |5 kcal mol⁻¹|) on each system are highlighted in Figure 5.

Figure 5.

Residues with significant contribution (ΔE = |5 kcal mol⁻¹|) to the stabilization of incoming nucleotide, terminal primer nucleotide, and two active site metals in Mg (blue) and Mn (Red). Numbers in parentheses indicate conservation score of each residue. 9 indicates highly conserved, and 1 indicates least conserved, and (–) indicates no data.

Residues R853, R866, R869, E895, K925, R943, E944, and E1192 show larger favorable intermolecular interactions with the active site in the presence of Mg²⁺, with residues R853 and K925 showing the largest interactions (21.9 kcal mol⁻¹ and 50.7 kcal mol⁻¹, respectively). For the system with Mn²⁺, T849, I850, E856, K875, K947, R1190, and K1191 show significant favorable intermolecular interactions with the active site, with no specific residue having a very large contribution when compared to the interactions in the Mg²⁺ system.

All the residues that have a significant contribution to the interactions with the active site are evolutionarily conserved, and some are sites for pathogenic mutations as well.^81,82 A previous study on a mutation of R853 (R853A),³⁰ which is linked to diseases such as progressive external opthalmoplegia (PEO) indicates that the mutation changes the structural and dynamic features of the protein, which could render it catalytically incompetent, providing a possible explanation for the observed experimental results. Other residues that are common mutagenesis sites are T849, E895, K925, R943, and K1191, leading to Alpers and other infantile hepatocerebral syndromes with mt DNA depletion, whereas R869 and R943 are also related to PEO.

3.2. Catalytic Effects of the Metals.

QM/MM investigations have been used in multiple studies related to polymerases to study the polymerization reaction of DNA.^13,23,44 The two steps of polymerization are the deprotonation of the O3′, followed by the nucleophilic attack/phosphoryl transfer, with the latter generally reported to be the rate-limiting step. In this study, we will focus on the second part of the reaction; that is, we will concentrate on the nucleophilic attack step after the deprotonation of the O3′.

The QM region for both systems includes D890, V891, D892, H1134, D1135, E1136, dT3′, dCTP, the two metal ions and coordinating water molecules to both metals, and the triphosphate part of dCTP. Five structures for each of the systems were selected as starting structures for reactant optimization based on catalytic competency. Once the QM/MM optimizations of the reactant were completed, the lowest energy representative was used in each system to generate the product structure in silico, which was subsequently optimized. Once the optimized product was obtained, QSM was utilized to obtain the minimum energy path for the reaction. Overlaying the reactant, approximate transition state, and product of each of the two systems shows that they have similar conformations (Figure S15)

As shown in Figure 6, the reactions catalyzed by Pol γ with either of the ions are exoergic, with the system containing Mn²⁺ ions having a lower reaction energy (ΔE = −3.6 kcal mol⁻¹) as compared to the system with Mg²⁺ (ΔE = −1.6 kcal mol⁻¹). The energy barrier for the system with Mn²⁺ ions is also lower (16.8 kcal mol⁻¹) than that of the system with Mg²⁺ ions (18.0 kcal mol⁻¹). Both systems have a pentacovalent phosphorane structure in the approximate transition state, with the O3′···Pα···O3α angle being 170.36° for Mn²⁺ ions and 169.24° for Mg²⁺ ions, which are favorable angles for S_N2 intermediate during the nucleotidyl transfer and consistent with other calculated TS structures for DNA addition by DNA polymerases.^13,23,44 The calculated barriers are consistent with experimental results,⁴⁹ where it was seen that the k_pol for catalytic pol γA in the presence of Mg²⁺ ions with correct dTTP incorporation is around 3.5 s⁻¹. Using the presteady state kinetics formula for the incorporation of a single nucleotide via the transition state formula $\left(k_{\text{pol}}=\frac{k_{\text{b}}T}{h}{\mathrm{e}}^{-E^{‡}/\mathit{\text{RT}}}\right)$, the experimental energy barrier was found to be around 16.7 kcal mol⁻¹. These results are consistent with other studies that carried out QM/MM studies on DNA polymerases.^{12–14,23,44,71,83–86}

Figure 6.

(A) Optimized geometries of the reactant, transition state and product for Mg²⁺ (above) and Mn²⁺ (below) systems. Critical distance and angle have also been mentioned. (B) Optimized QM/MM minimum energy path (kcal/mmol) for the catalytic reaction path.

EDA was also performed on the QM/MM optimized reactant, transition state, and product structures to analyze the effects of the protein environment on the reaction path. The difference in total intermolecular interaction between the Mn²⁺ and Mg²⁺ systems (QM subsystem vs the rest of the protein) for the reactant and TS is −25.5 and −3.2 kcal mol⁻¹, respectively. When comparing the interactions in the R → TS pathway, it is seen that $\Sigma E_{\text{NB}}^{\text{TS}-\text{Reactant}}$ value for the system with Mn²⁺ ions is −46.0 kcal mol⁻¹, whereas for the system with Mg²⁺ ions is −23.1 kcal mol⁻¹. This suggests that the protein environment stabilizes the active site in the reactant and transition state of both systems, with the system with Mn²⁺ ions having higher relative stabilization from the protein environment, which also helps explain the lower energy barrier.

Residues with considerable interactions (ΔE = |5 kcal mol⁻¹|) are shown in Table 1. As is seen from the table, most residues that have significant contributions are evolutionarily conserved, in particular, residues closer to the active site. Of the residues with favorable interactions in the reactant, residue Q894 has considerably higher favorable interaction in the reactant in both systems (23.6 kcal mol⁻¹ for the Mg²⁺ system and 12.7 kcal mol⁻¹ for the Mn²⁺ system). Among the residues with favorable interactions with the active site, stronger interactions are exhibited by G923 (13.8 kcal mol⁻¹) and R1190 (21.1 kcal mol⁻¹) in the system with Mg²⁺ ions, whereas residue A676 (−15.0 kcal mol⁻¹) shows a strong favorable interaction in the system with Mn²⁺ ions. It is interesting to note that G923 shows more favorable interactions toward the active site of the transition state in the presence of Mg²⁺ ions, whereas for Mn²⁺ ions, it shows more favorable interactions in the reactant.

Table 1.

Important Residues in with Considerable Stabilization (Red) or Destabilization (Blue) (ΔE = |5 kcal mol⁻¹|) of the Transition State as Compared to the Reactant^a

Residue	ΔETS-Reactant (kcal/mol)/MG	Residue	ΔETS-Reactant (kcal/mol)/MN
Q333(3)	−5.2	A676(4)	−15.0
R334(1)	−5.6	T682(4)	6.6
R852(9)	−6.5	N795(4)	5.2
Q894(9)	23.6	R852(9)	−5.2
G923(9)	−13.8	Q894(9)	12.7
T929(9)	5.6	G923(9)	6.5
V1137(9)	7.0	R924(8)	−7.5
T1160(9)	5.2	E980(−)	−8.2
R1190(9)	−21.1	T989(9)	−6.1
M1206(6)	−5.4	V1137(9)	8.2
		T1160(9)	6.7
		V1183(9)	−7.2
		R1190(9)	−5.9
		M1206(7)	−5.2
		L1218(8)	−9.4

^aNumbers in brackets indicate conservation score of each residue. 9 indicates highly conserved, 1 indicates least conserved, and (–) indicates no data.

Along with the R→TS path, when the R→P interactions are compared, it is seen that the active site in the product state is stabilized more in the system with Mn²⁺ $(\Sigma E_{\text{NB}}^{\text{Product}-\text{Reactant}}=-33.7\text{kcal}/\text{mol})$ ions when compared to the system with Mg²⁺ $(\Sigma E_{\text{NB}}^{\text{Product}-\text{Reactant}}=-19.3\text{kcal}/\text{mol})$ T1053 (22.4 kcal mol⁻¹) shows strong favorable interactions with the active site of the reactant in the system with Mg²⁺ ions whereas L1218 (18.0 kcal mol⁻¹) shows strong favorable interactions with the active site of the reactant in the system with Mn²⁺ ions. When looking at favorable interactions with the active site of the product, residues R852 (17.6 kcal mol⁻¹), R924 (22.7 kcal mol⁻¹), and T1160 (24.9 kcal mol⁻¹) show strong interactions in the system with Mg²⁺ ions, whereas residues A767 (21.6 kcal mol⁻¹), R924 (19.5 kcal mol⁻¹), T1160 (27.6 kcal mol⁻¹), and R1190 (33.9 kcal mol⁻¹) show strong interactions in the system with Mn²⁺ ions. It is interesting to note that most residues with favorable interactions with the product seem to be positively charged, which might indicate stabilization of the charge formed in the active site due to PPi. Other residues with considerable interactions (ΔE = |10 kcal mol⁻¹|) are shown in Table 2. Most of the residues that have considerable interactions in the reaction path are highly conserved evolutionarily, and most are frequent sites for mutagenesis that can cause various diseases.

Table 2.

Important Residues in with Considerable Stabilization (Red) or Destabilization (Blue) (ΔE = |10 kcal mol⁻¹|) of the Product as Compared to the Reactant^a

Residue	ΔEProduct-Reactant (kcal/mol)/MG	Residue	ΔEProduct-Reactant (kcal/mol)/MN
S36(9)	10.0	Q671(2)	11.9
Q39(1)	−11.2	T682(2)	10.6
V123(9)	14.9	V691(−)	11.1
A676(3)	10.9	A767(8)	−21.6
V691(−)	12.0	R852(9)	−14.7
K755(9)	10.0	A865(9)	−11.7
K768(9)	10.1	D868(8)	−11.2
P772(9)	−10.7	R924(8)	−19.5
N795(9)	−10.0	S933(9)	−14.9
T851(9)	−10.4	S942(9)	−10.6
R852(9)	−17.6	A946(9)	−13.5
A865(9)	−11.1	G952(9)	−11.2
D868(8)	−14.3	A962(9)	10.0
G903(9)	13.4	T1053(9)	12.8
G923(9)	−13.8	N1059(9)	−11.1
R924(8)	−22.7	V1106(8)	10.6
T929(9)	11.9	Y1147(−)	−12.7
S933(9)	−13.3	T1160(9)	−27.6
S942(9)	−10.8	N1171(2)	11.2
A946(9)	−11.5	R1190(9)	−33.9
A962(9)	11.4	G1215(9)	11.2
T989(9)	−12.1	L1218(7)	17.9
T1053(9)	22.4
T1160(9)	−25.0
N1171(2)	10.2
G1215(9)	10.4

^aNumbers in brackets indicate conservation score of each residue. 9 indicates highly conserved, 1 indicates least conserved, and (–) indicates no data.

The NCI index analysis for the reaction path shows favorable interactions throughout the reaction path. Looking at the interactions in the reactant, it is seen that the reactant with the Mn²⁺ ion shows stronger attractive interactions between the O3′ and Pα of the incoming nucleotide (shown by blue surfaces) when compared to the system with Mg²⁺ (shown in green, which signifies weak interactions) (Figure 7).

Figure 7.

Noncovalent interactions between dCTP and surrounding residues in the reactant of Mg²⁺ (left) and Mn²⁺ (right) systems. Zoomed in views for the O3′–Pa interactions are shown.

The nature of the electric fields in the active site was analyzed using a multipolar decomposition of the electron localization function (ELF) basins for selected atoms. Significant differences in basin structures and dipole norms were observed between the two systems^13,43,44 (Figure 8). Examining the dipole ELF basins for the reactant, transition state, and product structures, it is seen that the sum of the dipole moments for the basins between O3′ and the catalytic metal is larger in the Mn²⁺ system ($\Delta {\mu }_{\text{Reactant}(\text{Mn}-\text{Mg})}^{\Sigma (\mathrm{O}{3}^{\prime }-\text{Cat}-\text{Metal})}=0.79$, $\Delta {\mu }_{\text{TS}(\text{Mn}-\text{Mg})}^{\Sigma (\mathrm{O}{3}^{\prime }-\text{Cat}-\text{Metal})}=0.82$,$\Delta {\mu }_{\text{Product}(\text{Mn}-\text{Mg})}^{\Sigma (\mathrm{O}{3}^{\prime }-\text{Cat}-\text{Metal})}=0.60$), indicating greater polarization of O3′ in the system with Mn²⁺ ions, suggesting a facilitation of the deprotonation step and subsequent nucleophilic attack in the presence of Mn²⁺. Additionally, the dipole orientations suggest that the electric field in the Mn²⁺ system facilitates the reaction, as evidenced by the consistent dipole moment direction for V(O3′–Mn) in both the reactant and transition state—a pattern not observed in the Mg²⁺ system. Furthermore, in the transition state, the trispynaptic V(O3′–catalytic metal–Pα) basin exhibits a higher population and dipole moment in the Mn²⁺ system compared to Mg²⁺ ($\Delta {\mu }_{\text{TS}(\text{Mn}-\text{Mg})}^{\Sigma (\mathrm{O}{3}^{\prime }-\text{Cat}-\text{Metal}-\mathrm{P}\alpha )}=0.12$). In the product state, the dipole moment between C3′ and O3′ is significantly higher in the Mg²⁺ system than in Mn²⁺.

Figure 8.

Basins with the directions of dipole moments shown for reactant, approximate transition state and product for the system with V²⁺ ions (A–C) and the system with Mn²⁺ ions (D–F). The corresponding tables indicate the dipole moments of the important basins. The vectors shown in the figures are as follows: magenta: V(O3′–Cat metal), green (O3′–C3′), pink (O3′–Pα), blue(O3′–Cat Metal–Pα), and cyan (Pα–O3α).

Overall, the ELF analysis of the basin polarization is consistent with the reaction energetics and highlights the distinct influence of Mg²⁺ and Mn²⁺ on the active site’s electric field and its implications for reaction kinetics. Furthermore, basins with the nucleotide-binding metal show consistent populations with the three triphosphate nonbridging oxygens, with the basin with the α-oxygen increasing as the transition state approaches. This might indicate the stabilizing electric field provided by the nucleotide-binding metal with developing charges in the active site. Details about the populations of important basins along the reaction path, as well as specific basins related to the reactant, transition state, and product are shown in Tables S2–S5.

Overall, the changes seen in the noncovalent and covalent interactions show higher facilitation of the reaction in the Mn²⁺ system as compared to Mg²⁺. Along with this, the protein environment is seen to stabilize the active site of the TS and the product of the system with Mn²⁺ ions more as compared to Mg²⁺ ions. It is also determined that there is higher anticorrelated motion between the thumb and the fingers in the system with Mn²⁺ ions (the motion necessary for DNA extension) as compared to the Mg²⁺ ions. Overall, these factors together might explain the lower energy barrier and reaction energy of the system with Mn²⁺ ions, which is consistent with the experimental results.

4. CONCLUSIONS

In this study, we explored the effects of magnesium (Mg²⁺) and manganese (Mn²⁺) ions on the structural and catalytic properties of human polymerase gamma using molecular dynamics (MD) simulations and hybrid quantum mechanics/molecular mechanics (QM/MM) calculations. Our findings reveal that the presence of Mn²⁺ leads to increased overall flexibility in the protein, accompanied by alterations in the secondary structures of specific regions. Additionally, differences in the dynamic behavior between the two systems suggest that Mn²⁺ impacts the enzyme’s conformational landscape more significantly than Mg²⁺.

Analyzing the effect of the protein environment on the active site, we observed that the protein provides more favorable interactions to the active site in an Mn²⁺ solution. Our kinetic and thermodynamic analyses indicate that the catalytic reaction is more favorable in the presence of Mn²⁺, as reflected by the reaction energy values: −1.61 kcal mol⁻¹ for Mg²⁺ and −3.65 kcal mol⁻¹ for Mn²⁺. Additionally, the activation energy barrier was lower for Mn²⁺, indicating that Mn²⁺ facilitates a more efficient reaction pathway, which is also observed by the differences in intermolecular interactions and electric fields around the active site with the two different cations.

These findings suggest that Mn²⁺ enhances catalytic efficiency by reducing energetic barriers and favoring the progression of the reaction by enhanced stabilization of the TS and product states via polarization of the active site. Individual residue interaction analysis is consistent with previously reported experimental mutagenesis, and several of the residues observed as catalytically relevant are also sites involved in pathogenic mutations. This analysis also uncovered additional catalytically relevant amino acids that could be targets for subsequent experimental mutagenesis studies. Collectively, this study provides insights into the mechanistic role of metal ions in polymerase gamma function, with potential implications for understanding metal ion regulation in mitochondrial DNA replication.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jctc.5c00435.

MD simulations protocol; water around the incoming nucleotide; structural analysis of MD simulations; average RMSD with error bars of replicates of polymerase domains; average RMSF with error bars of replicates; average distance of active site metals with error bars of replicates (PDF)

Videos for Normal mode analysis, minimum energy path and combined ELF-NCI changes during the reaction (ZIP)

Data Availability Statement

Sample of the simulation data for both systems along with starting coordinates and parameters is available at 10.5281/zenodo.14953103. Further data additional to those present in the paper are provided in the Supporting Information, which is publicly available.